diff --git a/.github/workflows/run_cron_unit_evals.yml b/.github/workflows/run_cron_unit_evals.yml
index e7dcb04f77f3507aa861aee30e152850cb36b600..5f1d13b5b8311a67ebff207d1211acffbdff5d6e 100644
--- a/.github/workflows/run_cron_unit_evals.yml
+++ b/.github/workflows/run_cron_unit_evals.yml
@@ -13,6 +13,12 @@ on:
 jobs:
   cron_unit_evals:
     runs-on: namespace-profile-16x32-ubuntu-2204
+    strategy:
+      matrix:
+        model:
+        - anthropic/claude-sonnet-4-5-latest
+        - anthropic/claude-opus-4-5-latest
+      fail-fast: false
     steps:
     - name: steps::checkout_repo
       uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
@@ -49,6 +55,7 @@ jobs:
         OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
         GOOGLE_AI_API_KEY: ${{ secrets.GOOGLE_AI_API_KEY }}
         GOOGLE_CLOUD_PROJECT: ${{ secrets.GOOGLE_CLOUD_PROJECT }}
+        ZED_AGENT_MODEL: ${{ matrix.model }}
     - name: steps::cleanup_cargo_config
       if: always()
       run: |
diff --git a/Cargo.lock b/Cargo.lock
index 4b619bc4d1d90f817bba19cc42a4b43df46cce16..e398d20beaa0a8f7f381b4389bc5d17f310b12f0 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -21017,6 +21017,7 @@ dependencies = [
  "indexmap",
  "indoc",
  "serde",
+ "serde_json",
  "toml 0.8.23",
  "toml_edit 0.22.27",
 ]
diff --git a/tooling/xtask/Cargo.toml b/tooling/xtask/Cargo.toml
index 38c491dac668cf008aaee38acb30f870bcf09852..13179b2eb69ba9a63ba6be5784907b78bba1b9f2 100644
--- a/tooling/xtask/Cargo.toml
+++ b/tooling/xtask/Cargo.toml
@@ -18,5 +18,6 @@ toml.workspace = true
 indoc.workspace = true
 indexmap.workspace = true
 serde.workspace = true
+serde_json.workspace = true
 toml_edit.workspace = true
 gh-workflow.workspace = true
diff --git a/tooling/xtask/src/tasks/workflows/run_agent_evals.rs b/tooling/xtask/src/tasks/workflows/run_agent_evals.rs
index 220d3872f72326f42845622b5e3c61f4819f4550..34a7c6885db061191f5c3eac447838439708fbfe 100644
--- a/tooling/xtask/src/tasks/workflows/run_agent_evals.rs
+++ b/tooling/xtask/src/tasks/workflows/run_agent_evals.rs
@@ -1,4 +1,7 @@
-use gh_workflow::{Event, Expression, Job, Run, Schedule, Step, Use, Workflow, WorkflowDispatch};
+use gh_workflow::{
+    Event, Expression, Job, Run, Schedule, Step, Strategy, Use, Workflow, WorkflowDispatch,
+};
+use serde_json::json;
 
 use crate::tasks::workflows::{
     runners::{self, Platform},
@@ -114,7 +117,31 @@ fn cron_unit_evals() -> NamedJob {
         "#}))
     }
 
-    named::job(unit_evals(None).add_step(send_failure_to_slack()))
+    named::job(cron_unit_evals_job().add_step(send_failure_to_slack()))
+}
+
+const UNIT_EVAL_MODELS: &[&str] = &[
+    "anthropic/claude-sonnet-4-5-latest",
+    "anthropic/claude-opus-4-5-latest",
+];
+
+fn cron_unit_evals_job() -> Job {
+    let script_step = add_api_keys(steps::script("./script/run-unit-evals"))
+        .add_env(("ZED_AGENT_MODEL", "${{ matrix.model }}"));
+
+    Job::default()
+        .runs_on(runners::LINUX_DEFAULT)
+        .strategy(Strategy::default().fail_fast(false).matrix(json!({
+            "model": UNIT_EVAL_MODELS
+        })))
+        .add_step(steps::checkout_repo())
+        .add_step(steps::setup_cargo_config(Platform::Linux))
+        .add_step(steps::cache_rust_dependencies_namespace())
+        .map(steps::install_linux_dependencies)
+        .add_step(steps::cargo_install_nextest())
+        .add_step(steps::clear_target_dir_if_large(Platform::Linux))
+        .add_step(script_step)
+        .add_step(steps::cleanup_cargo_config(Platform::Linux))
 }
 
 fn unit_evals(commit: Option<&WorkflowInput>) -> Job {