1use anthropic::AnthropicModelMode;
2use anyhow::{Context as _, Result, anyhow};
3use chrono::{DateTime, Utc};
4use client::{Client, ModelRequestUsage, UserStore, zed_urls};
5use futures::{
6 AsyncBufReadExt, FutureExt, Stream, StreamExt, future::BoxFuture, stream::BoxStream,
7};
8use google_ai::GoogleModelMode;
9use gpui::{
10 AnyElement, AnyView, App, AsyncApp, Context, Entity, SemanticVersion, Subscription, Task,
11};
12use http_client::http::{HeaderMap, HeaderValue};
13use http_client::{AsyncBody, HttpClient, Method, Response, StatusCode};
14use language_model::{
15 AuthenticateError, LanguageModel, LanguageModelCacheConfiguration,
16 LanguageModelCompletionError, LanguageModelCompletionEvent, LanguageModelId, LanguageModelName,
17 LanguageModelProvider, LanguageModelProviderId, LanguageModelProviderName,
18 LanguageModelProviderState, LanguageModelProviderTosView, LanguageModelRequest,
19 LanguageModelToolChoice, LanguageModelToolSchemaFormat, LlmApiToken,
20 ModelRequestLimitReachedError, PaymentRequiredError, RateLimiter, RefreshLlmTokenListener,
21};
22use proto::Plan;
23use release_channel::AppVersion;
24use schemars::JsonSchema;
25use serde::{Deserialize, Serialize, de::DeserializeOwned};
26use settings::SettingsStore;
27use smol::io::{AsyncReadExt, BufReader};
28use std::pin::Pin;
29use std::str::FromStr as _;
30use std::sync::Arc;
31use std::time::Duration;
32use thiserror::Error;
33use ui::{TintColor, prelude::*};
34use util::{ResultExt as _, maybe};
35use zed_llm_client::{
36 CLIENT_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, CURRENT_PLAN_HEADER_NAME, CompletionBody,
37 CompletionRequestStatus, CountTokensBody, CountTokensResponse, EXPIRED_LLM_TOKEN_HEADER_NAME,
38 ListModelsResponse, MODEL_REQUESTS_RESOURCE_HEADER_VALUE,
39 SERVER_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, SUBSCRIPTION_LIMIT_RESOURCE_HEADER_NAME,
40 TOOL_USE_LIMIT_REACHED_HEADER_NAME, ZED_VERSION_HEADER_NAME,
41};
42
43use crate::provider::anthropic::{AnthropicEventMapper, count_anthropic_tokens, into_anthropic};
44use crate::provider::google::{GoogleEventMapper, into_google};
45use crate::provider::open_ai::{OpenAiEventMapper, count_open_ai_tokens, into_open_ai};
46
47const PROVIDER_ID: LanguageModelProviderId = language_model::ZED_CLOUD_PROVIDER_ID;
48const PROVIDER_NAME: LanguageModelProviderName = language_model::ZED_CLOUD_PROVIDER_NAME;
49
50#[derive(Default, Clone, Debug, PartialEq)]
51pub struct ZedDotDevSettings {
52 pub available_models: Vec<AvailableModel>,
53}
54
55#[derive(Clone, Debug, PartialEq, Serialize, Deserialize, JsonSchema)]
56#[serde(rename_all = "lowercase")]
57pub enum AvailableProvider {
58 Anthropic,
59 OpenAi,
60 Google,
61}
62
63#[derive(Clone, Debug, PartialEq, Serialize, Deserialize, JsonSchema)]
64pub struct AvailableModel {
65 /// The provider of the language model.
66 pub provider: AvailableProvider,
67 /// The model's name in the provider's API. e.g. claude-3-5-sonnet-20240620
68 pub name: String,
69 /// The name displayed in the UI, such as in the assistant panel model dropdown menu.
70 pub display_name: Option<String>,
71 /// The size of the context window, indicating the maximum number of tokens the model can process.
72 pub max_tokens: usize,
73 /// The maximum number of output tokens allowed by the model.
74 pub max_output_tokens: Option<u64>,
75 /// The maximum number of completion tokens allowed by the model (o1-* only)
76 pub max_completion_tokens: Option<u64>,
77 /// Override this model with a different Anthropic model for tool calls.
78 pub tool_override: Option<String>,
79 /// Indicates whether this custom model supports caching.
80 pub cache_configuration: Option<LanguageModelCacheConfiguration>,
81 /// The default temperature to use for this model.
82 pub default_temperature: Option<f32>,
83 /// Any extra beta headers to provide when using the model.
84 #[serde(default)]
85 pub extra_beta_headers: Vec<String>,
86 /// The model's mode (e.g. thinking)
87 pub mode: Option<ModelMode>,
88}
89
90#[derive(Default, Clone, Debug, PartialEq, Serialize, Deserialize, JsonSchema)]
91#[serde(tag = "type", rename_all = "lowercase")]
92pub enum ModelMode {
93 #[default]
94 Default,
95 Thinking {
96 /// The maximum number of tokens to use for reasoning. Must be lower than the model's `max_output_tokens`.
97 budget_tokens: Option<u32>,
98 },
99}
100
101impl From<ModelMode> for AnthropicModelMode {
102 fn from(value: ModelMode) -> Self {
103 match value {
104 ModelMode::Default => AnthropicModelMode::Default,
105 ModelMode::Thinking { budget_tokens } => AnthropicModelMode::Thinking { budget_tokens },
106 }
107 }
108}
109
110pub struct CloudLanguageModelProvider {
111 client: Arc<Client>,
112 state: gpui::Entity<State>,
113 _maintain_client_status: Task<()>,
114}
115
116pub struct State {
117 client: Arc<Client>,
118 llm_api_token: LlmApiToken,
119 user_store: Entity<UserStore>,
120 status: client::Status,
121 accept_terms_of_service_task: Option<Task<Result<()>>>,
122 models: Vec<Arc<zed_llm_client::LanguageModel>>,
123 default_model: Option<Arc<zed_llm_client::LanguageModel>>,
124 default_fast_model: Option<Arc<zed_llm_client::LanguageModel>>,
125 recommended_models: Vec<Arc<zed_llm_client::LanguageModel>>,
126 _fetch_models_task: Task<()>,
127 _settings_subscription: Subscription,
128 _llm_token_subscription: Subscription,
129}
130
131impl State {
132 fn new(
133 client: Arc<Client>,
134 user_store: Entity<UserStore>,
135 status: client::Status,
136 cx: &mut Context<Self>,
137 ) -> Self {
138 let refresh_llm_token_listener = RefreshLlmTokenListener::global(cx);
139
140 Self {
141 client: client.clone(),
142 llm_api_token: LlmApiToken::default(),
143 user_store,
144 status,
145 accept_terms_of_service_task: None,
146 models: Vec::new(),
147 default_model: None,
148 default_fast_model: None,
149 recommended_models: Vec::new(),
150 _fetch_models_task: cx.spawn(async move |this, cx| {
151 maybe!(async move {
152 let (client, llm_api_token) = this
153 .read_with(cx, |this, _cx| (client.clone(), this.llm_api_token.clone()))?;
154
155 loop {
156 let status = this.read_with(cx, |this, _cx| this.status)?;
157 if matches!(status, client::Status::Connected { .. }) {
158 break;
159 }
160
161 cx.background_executor()
162 .timer(Duration::from_millis(100))
163 .await;
164 }
165
166 let response = Self::fetch_models(client, llm_api_token).await?;
167 this.update(cx, |this, cx| {
168 this.update_models(response, cx);
169 })
170 })
171 .await
172 .context("failed to fetch Zed models")
173 .log_err();
174 }),
175 _settings_subscription: cx.observe_global::<SettingsStore>(|_, cx| {
176 cx.notify();
177 }),
178 _llm_token_subscription: cx.subscribe(
179 &refresh_llm_token_listener,
180 move |this, _listener, _event, cx| {
181 let client = this.client.clone();
182 let llm_api_token = this.llm_api_token.clone();
183 cx.spawn(async move |this, cx| {
184 llm_api_token.refresh(&client).await?;
185 let response = Self::fetch_models(client, llm_api_token).await?;
186 this.update(cx, |this, cx| {
187 this.update_models(response, cx);
188 })
189 })
190 .detach_and_log_err(cx);
191 },
192 ),
193 }
194 }
195
196 fn is_signed_out(&self) -> bool {
197 self.status.is_signed_out()
198 }
199
200 fn authenticate(&self, cx: &mut Context<Self>) -> Task<Result<()>> {
201 let client = self.client.clone();
202 cx.spawn(async move |state, cx| {
203 client
204 .authenticate_and_connect(true, &cx)
205 .await
206 .into_response()?;
207 state.update(cx, |_, cx| cx.notify())
208 })
209 }
210
211 fn has_accepted_terms_of_service(&self, cx: &App) -> bool {
212 self.user_store
213 .read(cx)
214 .current_user_has_accepted_terms()
215 .unwrap_or(false)
216 }
217
218 fn accept_terms_of_service(&mut self, cx: &mut Context<Self>) {
219 let user_store = self.user_store.clone();
220 self.accept_terms_of_service_task = Some(cx.spawn(async move |this, cx| {
221 let _ = user_store
222 .update(cx, |store, cx| store.accept_terms_of_service(cx))?
223 .await;
224 this.update(cx, |this, cx| {
225 this.accept_terms_of_service_task = None;
226 cx.notify()
227 })
228 }));
229 }
230
231 fn update_models(&mut self, response: ListModelsResponse, cx: &mut Context<Self>) {
232 let mut models = Vec::new();
233
234 for model in response.models {
235 models.push(Arc::new(model.clone()));
236
237 // Right now we represent thinking variants of models as separate models on the client,
238 // so we need to insert variants for any model that supports thinking.
239 if model.supports_thinking {
240 models.push(Arc::new(zed_llm_client::LanguageModel {
241 id: zed_llm_client::LanguageModelId(format!("{}-thinking", model.id).into()),
242 display_name: format!("{} Thinking", model.display_name),
243 ..model
244 }));
245 }
246 }
247
248 self.default_model = models
249 .iter()
250 .find(|model| model.id == response.default_model)
251 .cloned();
252 self.default_fast_model = models
253 .iter()
254 .find(|model| model.id == response.default_fast_model)
255 .cloned();
256 self.recommended_models = response
257 .recommended_models
258 .iter()
259 .filter_map(|id| models.iter().find(|model| &model.id == id))
260 .cloned()
261 .collect();
262 self.models = models;
263 cx.notify();
264 }
265
266 async fn fetch_models(
267 client: Arc<Client>,
268 llm_api_token: LlmApiToken,
269 ) -> Result<ListModelsResponse> {
270 let http_client = &client.http_client();
271 let token = llm_api_token.acquire(&client).await?;
272
273 let request = http_client::Request::builder()
274 .method(Method::GET)
275 .uri(http_client.build_zed_llm_url("/models", &[])?.as_ref())
276 .header("Authorization", format!("Bearer {token}"))
277 .body(AsyncBody::empty())?;
278 let mut response = http_client
279 .send(request)
280 .await
281 .context("failed to send list models request")?;
282
283 if response.status().is_success() {
284 let mut body = String::new();
285 response.body_mut().read_to_string(&mut body).await?;
286 return Ok(serde_json::from_str(&body)?);
287 } else {
288 let mut body = String::new();
289 response.body_mut().read_to_string(&mut body).await?;
290 anyhow::bail!(
291 "error listing models.\nStatus: {:?}\nBody: {body}",
292 response.status(),
293 );
294 }
295 }
296}
297
298impl CloudLanguageModelProvider {
299 pub fn new(user_store: Entity<UserStore>, client: Arc<Client>, cx: &mut App) -> Self {
300 let mut status_rx = client.status();
301 let status = *status_rx.borrow();
302
303 let state = cx.new(|cx| State::new(client.clone(), user_store.clone(), status, cx));
304
305 let state_ref = state.downgrade();
306 let maintain_client_status = cx.spawn(async move |cx| {
307 while let Some(status) = status_rx.next().await {
308 if let Some(this) = state_ref.upgrade() {
309 _ = this.update(cx, |this, cx| {
310 if this.status != status {
311 this.status = status;
312 cx.notify();
313 }
314 });
315 } else {
316 break;
317 }
318 }
319 });
320
321 Self {
322 client,
323 state: state.clone(),
324 _maintain_client_status: maintain_client_status,
325 }
326 }
327
328 fn create_language_model(
329 &self,
330 model: Arc<zed_llm_client::LanguageModel>,
331 llm_api_token: LlmApiToken,
332 ) -> Arc<dyn LanguageModel> {
333 Arc::new(CloudLanguageModel {
334 id: LanguageModelId(SharedString::from(model.id.0.clone())),
335 model,
336 llm_api_token: llm_api_token.clone(),
337 client: self.client.clone(),
338 request_limiter: RateLimiter::new(4),
339 })
340 }
341}
342
343impl LanguageModelProviderState for CloudLanguageModelProvider {
344 type ObservableEntity = State;
345
346 fn observable_entity(&self) -> Option<gpui::Entity<Self::ObservableEntity>> {
347 Some(self.state.clone())
348 }
349}
350
351impl LanguageModelProvider for CloudLanguageModelProvider {
352 fn id(&self) -> LanguageModelProviderId {
353 PROVIDER_ID
354 }
355
356 fn name(&self) -> LanguageModelProviderName {
357 PROVIDER_NAME
358 }
359
360 fn icon(&self) -> IconName {
361 IconName::AiZed
362 }
363
364 fn default_model(&self, cx: &App) -> Option<Arc<dyn LanguageModel>> {
365 let default_model = self.state.read(cx).default_model.clone()?;
366 let llm_api_token = self.state.read(cx).llm_api_token.clone();
367 Some(self.create_language_model(default_model, llm_api_token))
368 }
369
370 fn default_fast_model(&self, cx: &App) -> Option<Arc<dyn LanguageModel>> {
371 let default_fast_model = self.state.read(cx).default_fast_model.clone()?;
372 let llm_api_token = self.state.read(cx).llm_api_token.clone();
373 Some(self.create_language_model(default_fast_model, llm_api_token))
374 }
375
376 fn recommended_models(&self, cx: &App) -> Vec<Arc<dyn LanguageModel>> {
377 let llm_api_token = self.state.read(cx).llm_api_token.clone();
378 self.state
379 .read(cx)
380 .recommended_models
381 .iter()
382 .cloned()
383 .map(|model| self.create_language_model(model, llm_api_token.clone()))
384 .collect()
385 }
386
387 fn provided_models(&self, cx: &App) -> Vec<Arc<dyn LanguageModel>> {
388 let llm_api_token = self.state.read(cx).llm_api_token.clone();
389 self.state
390 .read(cx)
391 .models
392 .iter()
393 .cloned()
394 .map(|model| self.create_language_model(model, llm_api_token.clone()))
395 .collect()
396 }
397
398 fn is_authenticated(&self, cx: &App) -> bool {
399 let state = self.state.read(cx);
400 !state.is_signed_out() && state.has_accepted_terms_of_service(cx)
401 }
402
403 fn authenticate(&self, _cx: &mut App) -> Task<Result<(), AuthenticateError>> {
404 Task::ready(Ok(()))
405 }
406
407 fn configuration_view(&self, _: &mut Window, cx: &mut App) -> AnyView {
408 cx.new(|_| ConfigurationView::new(self.state.clone()))
409 .into()
410 }
411
412 fn must_accept_terms(&self, cx: &App) -> bool {
413 !self.state.read(cx).has_accepted_terms_of_service(cx)
414 }
415
416 fn render_accept_terms(
417 &self,
418 view: LanguageModelProviderTosView,
419 cx: &mut App,
420 ) -> Option<AnyElement> {
421 let state = self.state.read(cx);
422 if state.has_accepted_terms_of_service(cx) {
423 return None;
424 }
425 Some(
426 render_accept_terms(view, state.accept_terms_of_service_task.is_some(), {
427 let state = self.state.clone();
428 move |_window, cx| {
429 state.update(cx, |state, cx| state.accept_terms_of_service(cx));
430 }
431 })
432 .into_any_element(),
433 )
434 }
435
436 fn reset_credentials(&self, _cx: &mut App) -> Task<Result<()>> {
437 Task::ready(Ok(()))
438 }
439}
440
441fn render_accept_terms(
442 view_kind: LanguageModelProviderTosView,
443 accept_terms_of_service_in_progress: bool,
444 accept_terms_callback: impl Fn(&mut Window, &mut App) + 'static,
445) -> impl IntoElement {
446 let thread_fresh_start = matches!(view_kind, LanguageModelProviderTosView::ThreadFreshStart);
447 let thread_empty_state = matches!(view_kind, LanguageModelProviderTosView::ThreadEmptyState);
448
449 let terms_button = Button::new("terms_of_service", "Terms of Service")
450 .style(ButtonStyle::Subtle)
451 .icon(IconName::ArrowUpRight)
452 .icon_color(Color::Muted)
453 .icon_size(IconSize::XSmall)
454 .when(thread_empty_state, |this| this.label_size(LabelSize::Small))
455 .on_click(move |_, _window, cx| cx.open_url("https://zed.dev/terms-of-service"));
456
457 let button_container = h_flex().child(
458 Button::new("accept_terms", "I accept the Terms of Service")
459 .when(!thread_empty_state, |this| {
460 this.full_width()
461 .style(ButtonStyle::Tinted(TintColor::Accent))
462 .icon(IconName::Check)
463 .icon_position(IconPosition::Start)
464 .icon_size(IconSize::Small)
465 })
466 .when(thread_empty_state, |this| {
467 this.style(ButtonStyle::Tinted(TintColor::Warning))
468 .label_size(LabelSize::Small)
469 })
470 .disabled(accept_terms_of_service_in_progress)
471 .on_click(move |_, window, cx| (accept_terms_callback)(window, cx)),
472 );
473
474 if thread_empty_state {
475 h_flex()
476 .w_full()
477 .flex_wrap()
478 .justify_between()
479 .child(
480 h_flex()
481 .child(
482 Label::new("To start using Zed AI, please read and accept the")
483 .size(LabelSize::Small),
484 )
485 .child(terms_button),
486 )
487 .child(button_container)
488 } else {
489 v_flex()
490 .w_full()
491 .gap_2()
492 .child(
493 h_flex()
494 .flex_wrap()
495 .when(thread_fresh_start, |this| this.justify_center())
496 .child(Label::new(
497 "To start using Zed AI, please read and accept the",
498 ))
499 .child(terms_button),
500 )
501 .child({
502 match view_kind {
503 LanguageModelProviderTosView::PromptEditorPopup => {
504 button_container.w_full().justify_end()
505 }
506 LanguageModelProviderTosView::Configuration => {
507 button_container.w_full().justify_start()
508 }
509 LanguageModelProviderTosView::ThreadFreshStart => {
510 button_container.w_full().justify_center()
511 }
512 LanguageModelProviderTosView::ThreadEmptyState => div().w_0(),
513 }
514 })
515 }
516}
517
518pub struct CloudLanguageModel {
519 id: LanguageModelId,
520 model: Arc<zed_llm_client::LanguageModel>,
521 llm_api_token: LlmApiToken,
522 client: Arc<Client>,
523 request_limiter: RateLimiter,
524}
525
526struct PerformLlmCompletionResponse {
527 response: Response<AsyncBody>,
528 usage: Option<ModelRequestUsage>,
529 tool_use_limit_reached: bool,
530 includes_status_messages: bool,
531}
532
533impl CloudLanguageModel {
534 async fn perform_llm_completion(
535 client: Arc<Client>,
536 llm_api_token: LlmApiToken,
537 app_version: Option<SemanticVersion>,
538 body: CompletionBody,
539 ) -> Result<PerformLlmCompletionResponse> {
540 let http_client = &client.http_client();
541
542 let mut token = llm_api_token.acquire(&client).await?;
543 let mut refreshed_token = false;
544
545 loop {
546 let request_builder = http_client::Request::builder()
547 .method(Method::POST)
548 .uri(http_client.build_zed_llm_url("/completions", &[])?.as_ref());
549 let request_builder = if let Some(app_version) = app_version {
550 request_builder.header(ZED_VERSION_HEADER_NAME, app_version.to_string())
551 } else {
552 request_builder
553 };
554
555 let request = request_builder
556 .header("Content-Type", "application/json")
557 .header("Authorization", format!("Bearer {token}"))
558 .header(CLIENT_SUPPORTS_STATUS_MESSAGES_HEADER_NAME, "true")
559 .body(serde_json::to_string(&body)?.into())?;
560 let mut response = http_client.send(request).await?;
561 let status = response.status();
562 if status.is_success() {
563 let includes_status_messages = response
564 .headers()
565 .get(SERVER_SUPPORTS_STATUS_MESSAGES_HEADER_NAME)
566 .is_some();
567
568 let tool_use_limit_reached = response
569 .headers()
570 .get(TOOL_USE_LIMIT_REACHED_HEADER_NAME)
571 .is_some();
572
573 let usage = if includes_status_messages {
574 None
575 } else {
576 ModelRequestUsage::from_headers(response.headers()).ok()
577 };
578
579 return Ok(PerformLlmCompletionResponse {
580 response,
581 usage,
582 includes_status_messages,
583 tool_use_limit_reached,
584 });
585 }
586
587 if !refreshed_token
588 && response
589 .headers()
590 .get(EXPIRED_LLM_TOKEN_HEADER_NAME)
591 .is_some()
592 {
593 token = llm_api_token.refresh(&client).await?;
594 refreshed_token = true;
595 continue;
596 }
597
598 if status == StatusCode::FORBIDDEN
599 && response
600 .headers()
601 .get(SUBSCRIPTION_LIMIT_RESOURCE_HEADER_NAME)
602 .is_some()
603 {
604 if let Some(MODEL_REQUESTS_RESOURCE_HEADER_VALUE) = response
605 .headers()
606 .get(SUBSCRIPTION_LIMIT_RESOURCE_HEADER_NAME)
607 .and_then(|resource| resource.to_str().ok())
608 {
609 if let Some(plan) = response
610 .headers()
611 .get(CURRENT_PLAN_HEADER_NAME)
612 .and_then(|plan| plan.to_str().ok())
613 .and_then(|plan| zed_llm_client::Plan::from_str(plan).ok())
614 {
615 let plan = match plan {
616 zed_llm_client::Plan::ZedFree => Plan::Free,
617 zed_llm_client::Plan::ZedPro => Plan::ZedPro,
618 zed_llm_client::Plan::ZedProTrial => Plan::ZedProTrial,
619 };
620 return Err(anyhow!(ModelRequestLimitReachedError { plan }));
621 }
622 }
623 } else if status == StatusCode::PAYMENT_REQUIRED {
624 return Err(anyhow!(PaymentRequiredError));
625 }
626
627 let mut body = String::new();
628 let headers = response.headers().clone();
629 response.body_mut().read_to_string(&mut body).await?;
630 return Err(anyhow!(ApiError {
631 status,
632 body,
633 headers
634 }));
635 }
636 }
637}
638
639#[derive(Debug, Error)]
640#[error("cloud language model request failed with status {status}: {body}")]
641struct ApiError {
642 status: StatusCode,
643 body: String,
644 headers: HeaderMap<HeaderValue>,
645}
646
647impl From<ApiError> for LanguageModelCompletionError {
648 fn from(error: ApiError) -> Self {
649 let retry_after = None;
650 LanguageModelCompletionError::from_http_status(
651 PROVIDER_NAME,
652 error.status,
653 error.body,
654 retry_after,
655 )
656 }
657}
658
659impl LanguageModel for CloudLanguageModel {
660 fn id(&self) -> LanguageModelId {
661 self.id.clone()
662 }
663
664 fn name(&self) -> LanguageModelName {
665 LanguageModelName::from(self.model.display_name.clone())
666 }
667
668 fn provider_id(&self) -> LanguageModelProviderId {
669 PROVIDER_ID
670 }
671
672 fn provider_name(&self) -> LanguageModelProviderName {
673 PROVIDER_NAME
674 }
675
676 fn upstream_provider_id(&self) -> LanguageModelProviderId {
677 use zed_llm_client::LanguageModelProvider::*;
678 match self.model.provider {
679 Anthropic => language_model::ANTHROPIC_PROVIDER_ID,
680 OpenAi => language_model::OPEN_AI_PROVIDER_ID,
681 Google => language_model::GOOGLE_PROVIDER_ID,
682 }
683 }
684
685 fn upstream_provider_name(&self) -> LanguageModelProviderName {
686 use zed_llm_client::LanguageModelProvider::*;
687 match self.model.provider {
688 Anthropic => language_model::ANTHROPIC_PROVIDER_NAME,
689 OpenAi => language_model::OPEN_AI_PROVIDER_NAME,
690 Google => language_model::GOOGLE_PROVIDER_NAME,
691 }
692 }
693
694 fn supports_tools(&self) -> bool {
695 self.model.supports_tools
696 }
697
698 fn supports_images(&self) -> bool {
699 self.model.supports_images
700 }
701
702 fn supports_tool_choice(&self, choice: LanguageModelToolChoice) -> bool {
703 match choice {
704 LanguageModelToolChoice::Auto
705 | LanguageModelToolChoice::Any
706 | LanguageModelToolChoice::None => true,
707 }
708 }
709
710 fn supports_burn_mode(&self) -> bool {
711 self.model.supports_max_mode
712 }
713
714 fn telemetry_id(&self) -> String {
715 format!("zed.dev/{}", self.model.id)
716 }
717
718 fn tool_input_format(&self) -> LanguageModelToolSchemaFormat {
719 match self.model.provider {
720 zed_llm_client::LanguageModelProvider::Anthropic
721 | zed_llm_client::LanguageModelProvider::OpenAi => {
722 LanguageModelToolSchemaFormat::JsonSchema
723 }
724 zed_llm_client::LanguageModelProvider::Google => {
725 LanguageModelToolSchemaFormat::JsonSchemaSubset
726 }
727 }
728 }
729
730 fn max_token_count(&self) -> u64 {
731 self.model.max_token_count as u64
732 }
733
734 fn max_token_count_in_burn_mode(&self) -> Option<u64> {
735 self.model
736 .max_token_count_in_max_mode
737 .filter(|_| self.model.supports_max_mode)
738 .map(|max_token_count| max_token_count as u64)
739 }
740
741 fn cache_configuration(&self) -> Option<LanguageModelCacheConfiguration> {
742 match &self.model.provider {
743 zed_llm_client::LanguageModelProvider::Anthropic => {
744 Some(LanguageModelCacheConfiguration {
745 min_total_token: 2_048,
746 should_speculate: true,
747 max_cache_anchors: 4,
748 })
749 }
750 zed_llm_client::LanguageModelProvider::OpenAi
751 | zed_llm_client::LanguageModelProvider::Google => None,
752 }
753 }
754
755 fn count_tokens(
756 &self,
757 request: LanguageModelRequest,
758 cx: &App,
759 ) -> BoxFuture<'static, Result<u64>> {
760 match self.model.provider {
761 zed_llm_client::LanguageModelProvider::Anthropic => count_anthropic_tokens(request, cx),
762 zed_llm_client::LanguageModelProvider::OpenAi => {
763 let model = match open_ai::Model::from_id(&self.model.id.0) {
764 Ok(model) => model,
765 Err(err) => return async move { Err(anyhow!(err)) }.boxed(),
766 };
767 count_open_ai_tokens(request, model, cx)
768 }
769 zed_llm_client::LanguageModelProvider::Google => {
770 let client = self.client.clone();
771 let llm_api_token = self.llm_api_token.clone();
772 let model_id = self.model.id.to_string();
773 let generate_content_request =
774 into_google(request, model_id.clone(), GoogleModelMode::Default);
775 async move {
776 let http_client = &client.http_client();
777 let token = llm_api_token.acquire(&client).await?;
778
779 let request_body = CountTokensBody {
780 provider: zed_llm_client::LanguageModelProvider::Google,
781 model: model_id,
782 provider_request: serde_json::to_value(&google_ai::CountTokensRequest {
783 generate_content_request,
784 })?,
785 };
786 let request = http_client::Request::builder()
787 .method(Method::POST)
788 .uri(
789 http_client
790 .build_zed_llm_url("/count_tokens", &[])?
791 .as_ref(),
792 )
793 .header("Content-Type", "application/json")
794 .header("Authorization", format!("Bearer {token}"))
795 .body(serde_json::to_string(&request_body)?.into())?;
796 let mut response = http_client.send(request).await?;
797 let status = response.status();
798 let headers = response.headers().clone();
799 let mut response_body = String::new();
800 response
801 .body_mut()
802 .read_to_string(&mut response_body)
803 .await?;
804
805 if status.is_success() {
806 let response_body: CountTokensResponse =
807 serde_json::from_str(&response_body)?;
808
809 Ok(response_body.tokens as u64)
810 } else {
811 Err(anyhow!(ApiError {
812 status,
813 body: response_body,
814 headers
815 }))
816 }
817 }
818 .boxed()
819 }
820 }
821 }
822
823 fn stream_completion(
824 &self,
825 request: LanguageModelRequest,
826 cx: &AsyncApp,
827 ) -> BoxFuture<
828 'static,
829 Result<
830 BoxStream<'static, Result<LanguageModelCompletionEvent, LanguageModelCompletionError>>,
831 LanguageModelCompletionError,
832 >,
833 > {
834 let thread_id = request.thread_id.clone();
835 let prompt_id = request.prompt_id.clone();
836 let intent = request.intent;
837 let mode = request.mode;
838 let app_version = cx.update(|cx| AppVersion::global(cx)).ok();
839 let thinking_allowed = request.thinking_allowed;
840 match self.model.provider {
841 zed_llm_client::LanguageModelProvider::Anthropic => {
842 let request = into_anthropic(
843 request,
844 self.model.id.to_string(),
845 1.0,
846 self.model.max_output_tokens as u64,
847 if thinking_allowed && self.model.id.0.ends_with("-thinking") {
848 AnthropicModelMode::Thinking {
849 budget_tokens: Some(4_096),
850 }
851 } else {
852 AnthropicModelMode::Default
853 },
854 );
855 let client = self.client.clone();
856 let llm_api_token = self.llm_api_token.clone();
857 let future = self.request_limiter.stream(async move {
858 let PerformLlmCompletionResponse {
859 response,
860 usage,
861 includes_status_messages,
862 tool_use_limit_reached,
863 } = Self::perform_llm_completion(
864 client.clone(),
865 llm_api_token,
866 app_version,
867 CompletionBody {
868 thread_id,
869 prompt_id,
870 intent,
871 mode,
872 provider: zed_llm_client::LanguageModelProvider::Anthropic,
873 model: request.model.clone(),
874 provider_request: serde_json::to_value(&request)
875 .map_err(|e| anyhow!(e))?,
876 },
877 )
878 .await
879 .map_err(|err| match err.downcast::<ApiError>() {
880 Ok(api_err) => anyhow!(LanguageModelCompletionError::from(api_err)),
881 Err(err) => anyhow!(err),
882 })?;
883
884 let mut mapper = AnthropicEventMapper::new();
885 Ok(map_cloud_completion_events(
886 Box::pin(
887 response_lines(response, includes_status_messages)
888 .chain(usage_updated_event(usage))
889 .chain(tool_use_limit_reached_event(tool_use_limit_reached)),
890 ),
891 move |event| mapper.map_event(event),
892 ))
893 });
894 async move { Ok(future.await?.boxed()) }.boxed()
895 }
896 zed_llm_client::LanguageModelProvider::OpenAi => {
897 let client = self.client.clone();
898 let model = match open_ai::Model::from_id(&self.model.id.0) {
899 Ok(model) => model,
900 Err(err) => return async move { Err(anyhow!(err).into()) }.boxed(),
901 };
902 let request = into_open_ai(
903 request,
904 model.id(),
905 model.supports_parallel_tool_calls(),
906 None,
907 );
908 let llm_api_token = self.llm_api_token.clone();
909 let future = self.request_limiter.stream(async move {
910 let PerformLlmCompletionResponse {
911 response,
912 usage,
913 includes_status_messages,
914 tool_use_limit_reached,
915 } = Self::perform_llm_completion(
916 client.clone(),
917 llm_api_token,
918 app_version,
919 CompletionBody {
920 thread_id,
921 prompt_id,
922 intent,
923 mode,
924 provider: zed_llm_client::LanguageModelProvider::OpenAi,
925 model: request.model.clone(),
926 provider_request: serde_json::to_value(&request)
927 .map_err(|e| anyhow!(e))?,
928 },
929 )
930 .await?;
931
932 let mut mapper = OpenAiEventMapper::new();
933 Ok(map_cloud_completion_events(
934 Box::pin(
935 response_lines(response, includes_status_messages)
936 .chain(usage_updated_event(usage))
937 .chain(tool_use_limit_reached_event(tool_use_limit_reached)),
938 ),
939 move |event| mapper.map_event(event),
940 ))
941 });
942 async move { Ok(future.await?.boxed()) }.boxed()
943 }
944 zed_llm_client::LanguageModelProvider::Google => {
945 let client = self.client.clone();
946 let request =
947 into_google(request, self.model.id.to_string(), GoogleModelMode::Default);
948 let llm_api_token = self.llm_api_token.clone();
949 let future = self.request_limiter.stream(async move {
950 let PerformLlmCompletionResponse {
951 response,
952 usage,
953 includes_status_messages,
954 tool_use_limit_reached,
955 } = Self::perform_llm_completion(
956 client.clone(),
957 llm_api_token,
958 app_version,
959 CompletionBody {
960 thread_id,
961 prompt_id,
962 intent,
963 mode,
964 provider: zed_llm_client::LanguageModelProvider::Google,
965 model: request.model.model_id.clone(),
966 provider_request: serde_json::to_value(&request)
967 .map_err(|e| anyhow!(e))?,
968 },
969 )
970 .await?;
971
972 let mut mapper = GoogleEventMapper::new();
973 Ok(map_cloud_completion_events(
974 Box::pin(
975 response_lines(response, includes_status_messages)
976 .chain(usage_updated_event(usage))
977 .chain(tool_use_limit_reached_event(tool_use_limit_reached)),
978 ),
979 move |event| mapper.map_event(event),
980 ))
981 });
982 async move { Ok(future.await?.boxed()) }.boxed()
983 }
984 }
985 }
986}
987
988#[derive(Serialize, Deserialize)]
989#[serde(rename_all = "snake_case")]
990pub enum CloudCompletionEvent<T> {
991 Status(CompletionRequestStatus),
992 Event(T),
993}
994
995fn map_cloud_completion_events<T, F>(
996 stream: Pin<Box<dyn Stream<Item = Result<CloudCompletionEvent<T>>> + Send>>,
997 mut map_callback: F,
998) -> BoxStream<'static, Result<LanguageModelCompletionEvent, LanguageModelCompletionError>>
999where
1000 T: DeserializeOwned + 'static,
1001 F: FnMut(T) -> Vec<Result<LanguageModelCompletionEvent, LanguageModelCompletionError>>
1002 + Send
1003 + 'static,
1004{
1005 stream
1006 .flat_map(move |event| {
1007 futures::stream::iter(match event {
1008 Err(error) => {
1009 vec![Err(LanguageModelCompletionError::from(error))]
1010 }
1011 Ok(CloudCompletionEvent::Status(event)) => {
1012 vec![Ok(LanguageModelCompletionEvent::StatusUpdate(event))]
1013 }
1014 Ok(CloudCompletionEvent::Event(event)) => map_callback(event),
1015 })
1016 })
1017 .boxed()
1018}
1019
1020fn usage_updated_event<T>(
1021 usage: Option<ModelRequestUsage>,
1022) -> impl Stream<Item = Result<CloudCompletionEvent<T>>> {
1023 futures::stream::iter(usage.map(|usage| {
1024 Ok(CloudCompletionEvent::Status(
1025 CompletionRequestStatus::UsageUpdated {
1026 amount: usage.amount as usize,
1027 limit: usage.limit,
1028 },
1029 ))
1030 }))
1031}
1032
1033fn tool_use_limit_reached_event<T>(
1034 tool_use_limit_reached: bool,
1035) -> impl Stream<Item = Result<CloudCompletionEvent<T>>> {
1036 futures::stream::iter(tool_use_limit_reached.then(|| {
1037 Ok(CloudCompletionEvent::Status(
1038 CompletionRequestStatus::ToolUseLimitReached,
1039 ))
1040 }))
1041}
1042
1043fn response_lines<T: DeserializeOwned>(
1044 response: Response<AsyncBody>,
1045 includes_status_messages: bool,
1046) -> impl Stream<Item = Result<CloudCompletionEvent<T>>> {
1047 futures::stream::try_unfold(
1048 (String::new(), BufReader::new(response.into_body())),
1049 move |(mut line, mut body)| async move {
1050 match body.read_line(&mut line).await {
1051 Ok(0) => Ok(None),
1052 Ok(_) => {
1053 let event = if includes_status_messages {
1054 serde_json::from_str::<CloudCompletionEvent<T>>(&line)?
1055 } else {
1056 CloudCompletionEvent::Event(serde_json::from_str::<T>(&line)?)
1057 };
1058
1059 line.clear();
1060 Ok(Some((event, (line, body))))
1061 }
1062 Err(e) => Err(e.into()),
1063 }
1064 },
1065 )
1066}
1067
1068#[derive(IntoElement, RegisterComponent)]
1069struct ZedAiConfiguration {
1070 is_connected: bool,
1071 plan: Option<proto::Plan>,
1072 subscription_period: Option<(DateTime<Utc>, DateTime<Utc>)>,
1073 eligible_for_trial: bool,
1074 has_accepted_terms_of_service: bool,
1075 accept_terms_of_service_in_progress: bool,
1076 accept_terms_of_service_callback: Arc<dyn Fn(&mut Window, &mut App) + Send + Sync>,
1077 sign_in_callback: Arc<dyn Fn(&mut Window, &mut App) + Send + Sync>,
1078}
1079
1080impl RenderOnce for ZedAiConfiguration {
1081 fn render(self, _window: &mut Window, _cx: &mut App) -> impl IntoElement {
1082 const ZED_PRICING_URL: &str = "https://zed.dev/pricing";
1083
1084 let is_pro = self.plan == Some(proto::Plan::ZedPro);
1085 let subscription_text = match (self.plan, self.subscription_period) {
1086 (Some(proto::Plan::ZedPro), Some(_)) => {
1087 "You have access to Zed's hosted LLMs through your Zed Pro subscription."
1088 }
1089 (Some(proto::Plan::ZedProTrial), Some(_)) => {
1090 "You have access to Zed's hosted LLMs through your Zed Pro trial."
1091 }
1092 (Some(proto::Plan::Free), Some(_)) => {
1093 "You have basic access to Zed's hosted LLMs through your Zed Free subscription."
1094 }
1095 _ => {
1096 if self.eligible_for_trial {
1097 "Subscribe for access to Zed's hosted LLMs. Start with a 14 day free trial."
1098 } else {
1099 "Subscribe for access to Zed's hosted LLMs."
1100 }
1101 }
1102 };
1103 let manage_subscription_buttons = if is_pro {
1104 h_flex().child(
1105 Button::new("manage_settings", "Manage Subscription")
1106 .style(ButtonStyle::Tinted(TintColor::Accent))
1107 .on_click(|_, _, cx| cx.open_url(&zed_urls::account_url(cx))),
1108 )
1109 } else {
1110 h_flex()
1111 .gap_2()
1112 .child(
1113 Button::new("learn_more", "Learn more")
1114 .style(ButtonStyle::Subtle)
1115 .on_click(|_, _, cx| cx.open_url(ZED_PRICING_URL)),
1116 )
1117 .child(
1118 Button::new(
1119 "upgrade",
1120 if self.plan.is_none() && self.eligible_for_trial {
1121 "Start Trial"
1122 } else {
1123 "Upgrade"
1124 },
1125 )
1126 .style(ButtonStyle::Subtle)
1127 .color(Color::Accent)
1128 .on_click(|_, _, cx| cx.open_url(&zed_urls::account_url(cx))),
1129 )
1130 };
1131
1132 if self.is_connected {
1133 v_flex()
1134 .gap_3()
1135 .w_full()
1136 .when(!self.has_accepted_terms_of_service, |this| {
1137 this.child(render_accept_terms(
1138 LanguageModelProviderTosView::Configuration,
1139 self.accept_terms_of_service_in_progress,
1140 {
1141 let callback = self.accept_terms_of_service_callback.clone();
1142 move |window, cx| (callback)(window, cx)
1143 },
1144 ))
1145 })
1146 .when(self.has_accepted_terms_of_service, |this| {
1147 this.child(subscription_text)
1148 .child(manage_subscription_buttons)
1149 })
1150 } else {
1151 v_flex()
1152 .gap_2()
1153 .child(Label::new("Use Zed AI to access hosted language models."))
1154 .child(
1155 Button::new("sign_in", "Sign In")
1156 .icon_color(Color::Muted)
1157 .icon(IconName::Github)
1158 .icon_position(IconPosition::Start)
1159 .on_click({
1160 let callback = self.sign_in_callback.clone();
1161 move |_, window, cx| (callback)(window, cx)
1162 }),
1163 )
1164 }
1165 }
1166}
1167
1168struct ConfigurationView {
1169 state: Entity<State>,
1170 accept_terms_of_service_callback: Arc<dyn Fn(&mut Window, &mut App) + Send + Sync>,
1171 sign_in_callback: Arc<dyn Fn(&mut Window, &mut App) + Send + Sync>,
1172}
1173
1174impl ConfigurationView {
1175 fn new(state: Entity<State>) -> Self {
1176 let accept_terms_of_service_callback = Arc::new({
1177 let state = state.clone();
1178 move |_window: &mut Window, cx: &mut App| {
1179 state.update(cx, |state, cx| {
1180 state.accept_terms_of_service(cx);
1181 });
1182 }
1183 });
1184
1185 let sign_in_callback = Arc::new({
1186 let state = state.clone();
1187 move |_window: &mut Window, cx: &mut App| {
1188 state.update(cx, |state, cx| {
1189 state.authenticate(cx).detach_and_log_err(cx);
1190 });
1191 }
1192 });
1193
1194 Self {
1195 state,
1196 accept_terms_of_service_callback,
1197 sign_in_callback,
1198 }
1199 }
1200}
1201
1202impl Render for ConfigurationView {
1203 fn render(&mut self, _: &mut Window, cx: &mut Context<Self>) -> impl IntoElement {
1204 let state = self.state.read(cx);
1205 let user_store = state.user_store.read(cx);
1206
1207 ZedAiConfiguration {
1208 is_connected: !state.is_signed_out(),
1209 plan: user_store.current_plan(),
1210 subscription_period: user_store.subscription_period(),
1211 eligible_for_trial: user_store.trial_started_at().is_none(),
1212 has_accepted_terms_of_service: state.has_accepted_terms_of_service(cx),
1213 accept_terms_of_service_in_progress: state.accept_terms_of_service_task.is_some(),
1214 accept_terms_of_service_callback: self.accept_terms_of_service_callback.clone(),
1215 sign_in_callback: self.sign_in_callback.clone(),
1216 }
1217 }
1218}
1219
1220impl Component for ZedAiConfiguration {
1221 fn scope() -> ComponentScope {
1222 ComponentScope::Agent
1223 }
1224
1225 fn preview(_window: &mut Window, _cx: &mut App) -> Option<AnyElement> {
1226 fn configuration(
1227 is_connected: bool,
1228 plan: Option<proto::Plan>,
1229 eligible_for_trial: bool,
1230 has_accepted_terms_of_service: bool,
1231 ) -> AnyElement {
1232 ZedAiConfiguration {
1233 is_connected,
1234 plan,
1235 subscription_period: plan
1236 .is_some()
1237 .then(|| (Utc::now(), Utc::now() + chrono::Duration::days(7))),
1238 eligible_for_trial,
1239 has_accepted_terms_of_service,
1240 accept_terms_of_service_in_progress: false,
1241 accept_terms_of_service_callback: Arc::new(|_, _| {}),
1242 sign_in_callback: Arc::new(|_, _| {}),
1243 }
1244 .into_any_element()
1245 }
1246
1247 Some(
1248 v_flex()
1249 .p_4()
1250 .gap_4()
1251 .children(vec![
1252 single_example("Not connected", configuration(false, None, false, true)),
1253 single_example(
1254 "Accept Terms of Service",
1255 configuration(true, None, true, false),
1256 ),
1257 single_example(
1258 "No Plan - Not eligible for trial",
1259 configuration(true, None, false, true),
1260 ),
1261 single_example(
1262 "No Plan - Eligible for trial",
1263 configuration(true, None, true, true),
1264 ),
1265 single_example(
1266 "Free Plan",
1267 configuration(true, Some(proto::Plan::Free), true, true),
1268 ),
1269 single_example(
1270 "Zed Pro Trial Plan",
1271 configuration(true, Some(proto::Plan::ZedProTrial), true, true),
1272 ),
1273 single_example(
1274 "Zed Pro Plan",
1275 configuration(true, Some(proto::Plan::ZedPro), true, true),
1276 ),
1277 ])
1278 .into_any_element(),
1279 )
1280 }
1281}