Harm Weites and Andrey Nering created
Co-authored-by: Andrey Nering <andreynering@users.noreply.github.com>Taskfile.yaml | 5
cmd/nebius/main.go | 172 +++++++
internal/providers/configs/nebius.json | 653 +++++++++++++++++++++++++++
internal/providers/providers.go | 8
pkg/catwalk/provider.go | 2
5 files changed, 840 insertions(+)
@@ -99,6 +99,11 @@ tasks:
cmds:
- go run cmd/ionet/main.go
+ gen:nebius:
+ desc: Generate Nebius provider configurations
+ cmds:
+ - go run cmd/nebius/main.go
+
gen:openrouter:
desc: Generate openrouter provider configurations
cmds:
@@ -0,0 +1,172 @@
+// Package main provides a command-line tool to fetch models from Nebius Token Factory
+// and generate a configuration file for the provider.
+package main
+
+import (
+ "context"
+ "encoding/json"
+ "fmt"
+ "io"
+ "log"
+ "math"
+ "net/http"
+ "os"
+ "slices"
+ "strconv"
+ "strings"
+ "time"
+
+ "charm.land/catwalk/pkg/catwalk"
+)
+
+// Model represents a model from the Nebius Token Factory API.
+type Model struct {
+ ID string `json:"id"`
+ DisplayName string `json:"name"`
+ ContextLength int64 `json:"context_length"`
+ SupportedFeatures []string `json:"supported_features,omitempty"`
+ Pricing Pricing `json:"pricing"`
+ Architecture struct {
+ Modality string `json:"modality"`
+ } `json:"architecture,omitempty"`
+}
+
+type Pricing struct {
+ Prompt string `json:"prompt"`
+ Completion string `json:"completion"`
+}
+
+type ModelsResponse struct {
+ Data []Model `json:"data"`
+}
+
+func (m Model) hasFeature(featureValue string) bool {
+ if m.SupportedFeatures != nil {
+ for _, feature := range m.SupportedFeatures {
+ if strings.EqualFold(feature, featureValue) {
+ return true
+ }
+ }
+ }
+ return false
+}
+
+func fetchNebiusModels() (*ModelsResponse, error) {
+ client := &http.Client{Timeout: 30 * time.Second}
+ req, _ := http.NewRequestWithContext(
+ context.Background(),
+ "GET",
+ "https://api.tokenfactory.nebius.com/v1/models?verbose=true",
+ nil,
+ )
+ req.Header.Set("User-Agent", "Crush-Client/1.0")
+
+ // Read API key from environment variable
+ apiKey := os.Getenv("NEBIUS_API_KEY")
+ if apiKey == "" {
+ return nil, fmt.Errorf("$NEBIUS_API_KEY is required")
+ }
+ req.Header.Set("Authorization", "Bearer "+apiKey)
+
+ resp, err := client.Do(req)
+ if err != nil {
+ return nil, err //nolint:wrapcheck
+ }
+ defer resp.Body.Close() //nolint:errcheck
+ if resp.StatusCode != 200 {
+ body, _ := io.ReadAll(resp.Body)
+ return nil, fmt.Errorf("status %d: %s", resp.StatusCode, body)
+ }
+ var mr ModelsResponse
+ if err := json.NewDecoder(resp.Body).Decode(&mr); err != nil {
+ return nil, err //nolint:wrapcheck
+ }
+ return &mr, nil
+}
+
+func main() {
+ modelsResp, err := fetchNebiusModels()
+ if err != nil {
+ log.Fatal("Error fetching Nebius models:", err)
+ }
+
+ nebiusProvider := catwalk.Provider{
+ Name: "Nebius Token Factory",
+ ID: catwalk.InferenceProviderNebius,
+ APIKey: "$NEBIUS_API_KEY",
+ APIEndpoint: "https://api.tokenfactory.nebius.com/v1", // this is their default region, eu-north1
+ Type: catwalk.TypeOpenAICompat,
+ DefaultLargeModelID: "Qwen/Qwen3-Coder-30B-A3B-Instruct",
+ DefaultSmallModelID: "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B",
+ }
+
+ for _, model := range modelsResp.Data {
+ // we skip models that don't support tool calling
+ if !model.hasFeature("tools") {
+ continue
+ }
+
+ // Convert pricing from string to float64
+ var costPer1MIn, costPer1MOut float64
+
+ // Handle prompt price conversion
+ promptPrice, err := strconv.ParseFloat(model.Pricing.Prompt, 64)
+ if err != nil {
+ promptPrice = 0.0
+ }
+ costPer1MIn = math.Round(promptPrice*1_000_000*100) / 100 // Round to 2 decimal places
+
+ // Handle completion price conversion
+ completionPrice, err := strconv.ParseFloat(model.Pricing.Completion, 64)
+ if err != nil {
+ completionPrice = 0.0
+ }
+ costPer1MOut = math.Round(completionPrice*1_000_000*100) / 100 // Round to 2 decimal places
+
+ var (
+ supportsImages = strings.Contains(strings.ToLower(model.Architecture.Modality), "image")
+ canReason = model.hasFeature("reasoning")
+ reasoningLevels []string
+ defaultReasoning string
+ )
+ if canReason {
+ reasoningLevels = []string{"low", "medium", "high"}
+ defaultReasoning = "medium"
+ }
+
+ m := catwalk.Model{
+ ID: model.ID,
+ Name: model.DisplayName,
+ CostPer1MIn: costPer1MIn,
+ CostPer1MOut: costPer1MOut,
+ CostPer1MInCached: 0,
+ CostPer1MOutCached: 0,
+ ContextWindow: model.ContextLength,
+ DefaultMaxTokens: model.ContextLength / 10, // there is no MaxTokens exposed, so play safe
+ CanReason: canReason,
+ ReasoningLevels: reasoningLevels,
+ DefaultReasoningEffort: defaultReasoning,
+ SupportsImages: supportsImages,
+ }
+
+ nebiusProvider.Models = append(nebiusProvider.Models, m)
+ fmt.Printf("Added model %s with context window %d\n", model.ID, model.ContextLength)
+ }
+
+ slices.SortFunc(nebiusProvider.Models, func(a catwalk.Model, b catwalk.Model) int {
+ return strings.Compare(a.Name, b.Name)
+ })
+
+ // Save the JSON in internal/providers/configs/nebius.json
+ data, err := json.MarshalIndent(nebiusProvider, "", " ")
+ if err != nil {
+ log.Fatal("Error marshaling Nebius provider:", err)
+ }
+ data = append(data, '\n')
+
+ if err := os.WriteFile("internal/providers/configs/nebius.json", data, 0o600); err != nil {
+ log.Fatal("Error writing Nebius provider config:", err)
+ }
+
+ fmt.Printf("Generated nebius.json with %d models\n", len(nebiusProvider.Models))
+}
@@ -0,0 +1,653 @@
+{
+ "name": "Nebius Token Factory",
+ "id": "nebius",
+ "api_key": "$NEBIUS_API_KEY",
+ "api_endpoint": "https://api.tokenfactory.nebius.com/v1",
+ "type": "openai-compat",
+ "default_large_model_id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
+ "default_small_model_id": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B",
+ "models": [
+ {
+ "id": "deepseek-ai/DeepSeek-R1-0528",
+ "name": "DeepSeek-R1-0528",
+ "cost_per_1m_in": 0.8,
+ "cost_per_1m_out": 2.4,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 163840,
+ "default_max_tokens": 16384,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "deepseek-ai/DeepSeek-V3-0324",
+ "name": "DeepSeek-V3-0324",
+ "cost_per_1m_in": 0.5,
+ "cost_per_1m_out": 1.5,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 163840,
+ "default_max_tokens": 16384,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "deepseek-ai/DeepSeek-V3.2",
+ "name": "DeepSeek-V3.2",
+ "cost_per_1m_in": 0.3,
+ "cost_per_1m_out": 0.45,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 163000,
+ "default_max_tokens": 16300,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "deepseek-ai/DeepSeek-V3.2-fast",
+ "name": "DeepSeek-V3.2 (fast)",
+ "cost_per_1m_in": 0.4,
+ "cost_per_1m_out": 2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 8000,
+ "default_max_tokens": 800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "zai-org/GLM-4.5",
+ "name": "GLM-4.5",
+ "cost_per_1m_in": 0.6,
+ "cost_per_1m_out": 2.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "zai-org/GLM-4.5-Air",
+ "name": "GLM-4.5-Air",
+ "cost_per_1m_in": 0.2,
+ "cost_per_1m_out": 1.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "zai-org/GLM-4.7-FP8",
+ "name": "GLM-4.7",
+ "cost_per_1m_in": 0.4,
+ "cost_per_1m_out": 2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 202752,
+ "default_max_tokens": 20275,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "zai-org/GLM-5",
+ "name": "GLM-5",
+ "cost_per_1m_in": 1,
+ "cost_per_1m_out": 3.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 202752,
+ "default_max_tokens": 20275,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "NousResearch/Hermes-4-405B",
+ "name": "Hermes-4-405B",
+ "cost_per_1m_in": 1,
+ "cost_per_1m_out": 3,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "NousResearch/Hermes-4-70B",
+ "name": "Hermes-4-70B",
+ "cost_per_1m_in": 0.13,
+ "cost_per_1m_out": 0.4,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "PrimeIntellect/INTELLECT-3",
+ "name": "INTELLECT-3",
+ "cost_per_1m_in": 0.2,
+ "cost_per_1m_out": 1.1,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "moonshotai/Kimi-K2-Instruct",
+ "name": "Kimi-K2-Instruct",
+ "cost_per_1m_in": 0.5,
+ "cost_per_1m_out": 2.4,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "moonshotai/Kimi-K2-Thinking",
+ "name": "Kimi-K2-Thinking",
+ "cost_per_1m_in": 0.6,
+ "cost_per_1m_out": 2.5,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "moonshotai/Kimi-K2.5",
+ "name": "Kimi-K2.5",
+ "cost_per_1m_in": 0.5,
+ "cost_per_1m_out": 2.5,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "moonshotai/Kimi-K2.5-fast",
+ "name": "Kimi-K2.5 (fast)",
+ "cost_per_1m_in": 0.5,
+ "cost_per_1m_out": 2.5,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 8000,
+ "default_max_tokens": 800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "meta-llama/Llama-3.3-70B-Instruct",
+ "name": "Llama-3.3-70B-Instruct",
+ "cost_per_1m_in": 0.13,
+ "cost_per_1m_out": 0.4,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "meta-llama/Llama-3.3-70B-Instruct-fast",
+ "name": "Llama-3.3-70B-Instruct (fast)",
+ "cost_per_1m_in": 0.25,
+ "cost_per_1m_out": 0.75,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "meta-llama/Meta-Llama-3.1-8B-Instruct",
+ "name": "Meta-Llama-3.1-8B-Instruct",
+ "cost_per_1m_in": 0.02,
+ "cost_per_1m_out": 0.06,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "meta-llama/Meta-Llama-3.1-8B-Instruct-fast",
+ "name": "Meta-Llama-3.1-8B-Instruct (fast)",
+ "cost_per_1m_in": 0.03,
+ "cost_per_1m_out": 0.09,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "MiniMaxAI/MiniMax-M2.1",
+ "name": "MiniMax-M2.1",
+ "cost_per_1m_in": 0.3,
+ "cost_per_1m_out": 1.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 196608,
+ "default_max_tokens": 19660,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "MiniMaxAI/MiniMax-M2.5",
+ "name": "MiniMax-M2.5",
+ "cost_per_1m_in": 0.3,
+ "cost_per_1m_out": 1.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 196608,
+ "default_max_tokens": 19660,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B",
+ "name": "Nemotron-3-Nano-30B-A3B",
+ "cost_per_1m_in": 0.06,
+ "cost_per_1m_out": 0.24,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "nvidia/nemotron-3-super-120b-a12b",
+ "name": "Nemotron-3-Super-120b-a12b",
+ "cost_per_1m_in": 0.3,
+ "cost_per_1m_out": 0.9,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
+ "name": "Qwen3-235B-A22B-Instruct-2507",
+ "cost_per_1m_in": 0.2,
+ "cost_per_1m_out": 0.6,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
+ "name": "Qwen3-235B-A22B-Thinking-2507",
+ "cost_per_1m_in": 0.2,
+ "cost_per_1m_out": 0.8,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-235B-A22B-Thinking-2507-fast",
+ "name": "Qwen3-235B-A22B-Thinking-2507 (fast)",
+ "cost_per_1m_in": 0.5,
+ "cost_per_1m_out": 2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 8000,
+ "default_max_tokens": 800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
+ "name": "Qwen3-30B-A3B-Instruct-2507",
+ "cost_per_1m_in": 0.1,
+ "cost_per_1m_out": 0.3,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-30B-A3B-Thinking-2507",
+ "name": "Qwen3-30B-A3B-Thinking-2507",
+ "cost_per_1m_in": 0.1,
+ "cost_per_1m_out": 0.3,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-32B",
+ "name": "Qwen3-32B",
+ "cost_per_1m_in": 0.1,
+ "cost_per_1m_out": 0.3,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 40960,
+ "default_max_tokens": 4096,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-32B-fast",
+ "name": "Qwen3-32B (fast)",
+ "cost_per_1m_in": 0.2,
+ "cost_per_1m_out": 0.6,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 40960,
+ "default_max_tokens": 4096,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
+ "name": "Qwen3-Coder-30B-A3B-Instruct",
+ "cost_per_1m_in": 0.1,
+ "cost_per_1m_out": 0.3,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-Coder-480B-A35B-Instruct",
+ "name": "Qwen3-Coder-480B-A35B-Instruct",
+ "cost_per_1m_in": 0.4,
+ "cost_per_1m_out": 1.8,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": false,
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-Next-80B-A3B-Thinking",
+ "name": "Qwen3-Next-80B-A3B-Thinking",
+ "cost_per_1m_in": 0.15,
+ "cost_per_1m_out": 1.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 128000,
+ "default_max_tokens": 12800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3-Next-80B-A3B-Thinking-fast",
+ "name": "Qwen3-Next-80B-A3B-Thinking (fast)",
+ "cost_per_1m_in": 0.15,
+ "cost_per_1m_out": 1.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 8000,
+ "default_max_tokens": 800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3.5-397B-A17B",
+ "name": "Qwen3.5-397B-A17B",
+ "cost_per_1m_in": 0.6,
+ "cost_per_1m_out": 3.6,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 262144,
+ "default_max_tokens": 26214,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "Qwen/Qwen3.5-397B-A17B-fast",
+ "name": "Qwen3.5-397B-A17B (fast)",
+ "cost_per_1m_in": 0.6,
+ "cost_per_1m_out": 3.6,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 8000,
+ "default_max_tokens": 800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "openai/gpt-oss-120b",
+ "name": "gpt-oss-120b",
+ "cost_per_1m_in": 0.15,
+ "cost_per_1m_out": 0.6,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "openai/gpt-oss-120b-fast",
+ "name": "gpt-oss-120b (fast)",
+ "cost_per_1m_in": 0.1,
+ "cost_per_1m_out": 0.5,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 8000,
+ "default_max_tokens": 800,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ },
+ {
+ "id": "openai/gpt-oss-20b",
+ "name": "gpt-oss-20b",
+ "cost_per_1m_in": 0.05,
+ "cost_per_1m_out": 0.2,
+ "cost_per_1m_in_cached": 0,
+ "cost_per_1m_out_cached": 0,
+ "context_window": 131072,
+ "default_max_tokens": 13107,
+ "can_reason": true,
+ "reasoning_levels": [
+ "low",
+ "medium",
+ "high"
+ ],
+ "default_reasoning_effort": "medium",
+ "supports_attachments": false
+ }
+ ]
+}
@@ -42,6 +42,9 @@ var zhipuConfig []byte
//go:embed configs/zhipu-coding.json
var zhipuCodingConfig []byte
+//go:embed configs/nebius.json
+var nebiusConfig []byte
+
//go:embed configs/bedrock.json
var bedrockConfig []byte
@@ -125,6 +128,7 @@ var providerRegistry = []ProviderFunc{
ioNetProvider,
qiniuCloudProvider,
avianProvider,
+ nebiusProvider,
}
// GetAll returns all registered providers.
@@ -253,6 +257,10 @@ func qiniuCloudProvider() catwalk.Provider {
return loadProviderFromConfig(qiniuCloudConfig)
}
+func nebiusProvider() catwalk.Provider {
+ return loadProviderFromConfig(nebiusConfig)
+}
+
func avianProvider() catwalk.Provider {
return loadProviderFromConfig(avianConfig)
}
@@ -48,6 +48,7 @@ const (
InferenceProviderIoNet InferenceProvider = "ionet"
InferenceProviderQiniuCloud InferenceProvider = "qiniucloud"
InferenceProviderAvian InferenceProvider = "avian"
+ InferenceProviderNebius InferenceProvider = "nebius"
)
// Provider represents an AI provider configuration.
@@ -119,6 +120,7 @@ func KnownProviders() []InferenceProvider {
InferenceProviderMiniMaxChina,
InferenceProviderQiniuCloud,
InferenceProviderAvian,
+ InferenceProviderNebius,
}
}