Skip to content

A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.

Update LLM model catalog (#10571)

github-actions[bot]09c2c306f9ce
diff --git a/packages/llm-info/data/models.yml b/packages/llm-info/data/models.yml
index c7a1a6c801..b1a401b89e 100644
--- a/packages/llm-info/data/models.yml
+++ b/packages/llm-info/data/models.yml
@@ -211,6 +211,16 @@ openai:
release_date: 2025-12-16
google:
+ - name: Gemini 3.7 Flash
+ model: gemini-3.7-flash
+ description: "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning"
+ roles: [chat, edit]
+ capabilities: [thinking, tool_calling]
+ input_types: [text, image, pdf]
+ output_types: [text]
+ release_date: 2026-08-13
+ cost: {input: 0.75, output: 3.75}
+
- name: Gemini 3.6 Flash
model: gemini-3.6-flash
description: "Google's high-efficiency model for coding, agentic workflows, and web and app development"
@@ -920,6 +930,16 @@ ollama:
release_date: 2026-02-15
wandb:
+ - name: Nemotron 3.5 Lightning
+ model: nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B
+ description: "MoE model built for fast, reliable agentic tasks across financial services, cybersecurity, telecom, and retail"
+ roles: [chat, edit]
+ capabilities: [thinking, tool_calling]
+ input_types: [text]
+ output_types: [text]
+ release_date: 2026-08-11
+ cost: {input: 0.1, output: 0.25}
+
- name: DeepSeek V4 Flash 0731
model: deepseek-ai/DeepSeek-V4-Flash-0731
description: "DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total"
@@ -970,7 +990,7 @@ wandb:
release_date: 2026-04-24
cost: {input: 0.14, output: 0.28}
- - name: DeepSeek V4-Pro
+ - name: DeepSeek V4 Pro
model: deepseek-ai/DeepSeek-V4-Pro
description: "DeepSeek V4-Pro is a 1.6T-parameter MoE model with 49B active parameters excelling at advanced reasoning, coding, and complex agentic workloads"
roles: [chat, edit]
@@ -978,17 +998,17 @@ wandb:
input_types: [text]
output_types: [text]
release_date: 2026-04-24
- cost: {input: 1.74, output: 3.46}
+ cost: {input: 1.15, output: 2.55}
- - name: Z.AI GLM 5.2
+ - name: GLM 5.2
model: zai-org/GLM-5.2
description: "GLM-5.2 is a Mixture-of-Experts language model featuring 40 billion activated parameters and a total of 744 billion parameters"
roles: [chat, edit]
capabilities: [thinking, tool_calling]
input_types: [text]
output_types: [text]
- release_date: 2026-06-13
- cost: {input: 1.39, output: 4.4}
+ release_date: 2026-06-16
+ cost: {input: 0.76, output: 2.42}
- name: Z.AI GLM-5.1
model: zai-org/GLM-5.1
@@ -1030,7 +1050,7 @@ wandb:
release_date: 2026-02-12
cost: {input: 0.3, output: 1.2}
- - name: Moonshot AI Kimi K2.7 Code
+ - name: Kimi K2.7 Code
model: moonshotai/Kimi-K2.7-Code
description: "Kimi K2.7 Code is a 1T-parameter MoE model with 32B active parameters purpose-built for long-horizon agentic coding and software engineering"
roles: [chat, edit]
@@ -1038,7 +1058,7 @@ wandb:
input_types: [text]
output_types: [text]
release_date: 2026-06-12
- cost: {input: 0.94, output: 4}
+ cost: {input: 0.71, output: 3.5}
- name: Moonshot AI Kimi K2.6
model: moonshotai/Kimi-K2.6
@@ -1090,7 +1110,7 @@ wandb:
release_date: 2026-04-22
cost: {input: 0.6, output: 3.6}
- - name: JetBrains Mellum2 12B A2.5B
+ - name: Mellum2 12B A2.5B
model: JetBrains/Mellum2-12B-A2.5B-Instruct
description: "Mellum2-12B-A2.5B-Instruct is a fast MoE model with 131K context built for coding, tool use, and low-latency AI workflows"
roles: [chat, edit]
@@ -1101,6 +1121,16 @@ wandb:
cost: {input: 0.05, output: 0.1}
opencode-go:
+ - name: GLM-5.3
+ model: glm-5.3
+ description: "Flagship GLM model for long-horizon coding, agents, and complex project delivery"
+ roles: [chat, edit]
+ capabilities: [thinking, tool_calling]
+ input_types: [text]
+ output_types: [text]
+ release_date: 2026-08-14
+ cost: {input: 1.4, output: 4.4}
+
- name: Qwen3.8 Max
model: qwen3.8-max
description: "Flagship multimodal reasoning model in Alibaba's Qwen3.8 series, intended for complex reasoning, visual understanding, and agentic workflows"
diff --git a/packages/llm-info/src/__tests__/sync-models.test.ts b/packages/llm-info/src/__tests__/sync-models.test.ts
index 46df242e1e..58d4b31111 100644
--- a/packages/llm-info/src/__tests__/sync-models.test.ts
+++ b/packages/llm-info/src/__tests__/sync-models.test.ts
@@ -562,10 +562,19 @@ describe("mergeModels", () => {
release_date: date,
});
- it("seeds the denylist with gpt-5.3-codex-spark under openai", () => {
+ it("seeds the denylist with unsupported and redundant models", () => {
// Changing this set is a deliberate product decision — when a model is
// added or removed here, leave a comment next to it explaining why.
expect(MODEL_DENYLIST["openai"]?.has("gpt-5.3-codex-spark")).toBe(true);
+ expect(MODEL_DENYLIST["openai"]?.has("gpt-realtime-2.1")).toBe(true);
+ expect(MODEL_DENYLIST["bedrock"]).toEqual(
+ new Set([
+ "anthropic.claude-opus-5",
+ "au.anthropic.claude-opus-5",
+ "eu.anthropic.claude-opus-5",
+ "jp.anthropic.claude-opus-5",
+ ]),
+ );
});
it("drops upstream models that match the per-provider denylist", () => {
diff --git a/packages/llm-info/src/sources/merge.ts b/packages/llm-info/src/sources/merge.ts
index 0f311533ff..aeca0d1039 100644
--- a/packages/llm-info/src/sources/merge.ts
+++ b/packages/llm-info/src/sources/merge.ts
@@ -48,6 +48,15 @@ export const MODEL_DENYLIST: Readonly<Record<string, ReadonlySet<string>>> = {
openai: new Set([
// Surfaced inside ChatGPT only; not exposed via the OpenAI API.
"gpt-5.3-codex-spark",
+ // Listed upstream but does not work with marimo's OpenAI integration.
+ "gpt-realtime-2.1",
+ ]),
+ bedrock: new Set([
+ // Provider-specific aliases are redundant with the global catalog entry.
+ "anthropic.claude-opus-5",
+ "au.anthropic.claude-opus-5",
+ "eu.anthropic.claude-opus-5",
+ "jp.anthropic.claude-opus-5",
]),
};