improvement(models): add DeepSeek V4 + Mistral Medium 3.5, fix Codestral context window

[waleedlatif1]

Summary

• Add deepseek-v4-pro and deepseek-v4-flash (1M context, verified against DeepSeek pricing docs)
• Add mistral-medium-2604 (Mistral Medium 3.5, 256K context, $1.50/$7.50, verified against Mistral docs)
• Fix codestral-latest and codestral-2508 context window 128K → 256K (verified: Codestral 25.08 is 256K)

All values verified against each provider's live docs — cross-checked models.dev but did not treat it as source of truth (it had stale data for gpt-5-chat-latest, Ministral, and several -latest aliases, which were left unchanged because ours were already correct).

Type of Change

• Improvement (data accuracy)

Testing

Tested manually — models.ts parses cleanly and passes file-scoped typecheck.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Jun 17, 2026 12:55am

[cursor]

PR Summary

Low Risk
Declarative catalog-only changes; no runtime logic, auth, or API behavior—only model picker, pricing, and capability metadata.

Overview
Updates the provider model catalog in models.ts with new DeepSeek and Mistral entries and corrects Codestral context limits.

DeepSeek adds deepseek-v4-pro and deepseek-v4-flash ahead of existing chat models, each with 1M context, provider-aligned pricing (including cached input where applicable), and release metadata; flash includes temperature support.

Mistral adds mistral-medium-2604 (Medium 3.5) with 256K context and $1.50 / $7.50 input/output pricing.

Codestral bumps contextWindow from 128K to 256K for codestral-latest and codestral-2508 only; pricing and capabilities are unchanged.

^{Reviewed by Cursor Bugbot for commit 06472b5. Configure here.}

[greptile-apps]

Greptile Summary

This PR is a data accuracy update to models.ts, adding two new DeepSeek V4 models and one Mistral model, and correcting two Codestral context window values that were stale at 128K.

• DeepSeek: Adds deepseek-v4-pro (1M context, $0.435/$0.87, $0.003625 cached) and deepseek-v4-flash (1M context, $0.14/$0.28, $0.0028 cached) — pricing verified exactly against api-docs.deepseek.com. deepseek-v4-pro intentionally omits a temperature capability, matching the existing pattern for reasoning-focused models (deepseek-r1, deepseek-reasoner).
• Mistral: Adds mistral-medium-2604 (Mistral Medium 3.5, 256K context, $1.50/$7.50) inserted correctly before the older mistral-medium-2508 entry.
• Codestral fix: Bumps codestral-latest and codestral-2508 context window from 128K → 256K, confirmed by the Codestral 25.08 release announcement.

Confidence Score: 5/5

Safe to merge — all changed values are data-only corrections with no logic or API surface changes.

Every pricing value, model ID, and context window in the diff was cross-checked against authoritative sources: DeepSeek's official API docs confirm the V4-Pro and V4-Flash pricing to the cent (including the $0.003625 cached-input figure), the Codestral 25.08 announcement confirms 256K context, and the Mistral model naming convention and 256K context for Medium 3.5 are consistent with Mistral's docs. The temperature capability asymmetry between the two new DeepSeek models follows the established codebase pattern (reasoning-focused models carry empty capabilities objects).

No files require special attention.

Important Files Changed

Filename	Overview
apps/sim/providers/models.ts	Adds deepseek-v4-pro and deepseek-v4-flash (pricing confirmed against official DeepSeek API docs), mistral-medium-2604 (Mistral Medium 3.5), and corrects codestral-latest/codestral-2508 context window from 128K to 256K.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    subgraph DeepSeek["DeepSeek Provider"]
        DVP["deepseek-v4-pro ✨\n1M ctx | $0.435/$0.87\ncached $0.003625\nNo temperature cap"]
        DVF["deepseek-v4-flash ✨\n1M ctx | $0.14/$0.28\ncached $0.0028\ntemp 0–2"]
        DC["deepseek-chat (existing)\n1M ctx | $0.14/$0.28"]
        DR["deepseek-reasoner (existing)\n1M ctx | $0.14/$0.28"]
    end

    subgraph Mistral["Mistral Provider"]
        MM2604["mistral-medium-2604 ✨\n256K ctx | $1.50/$7.50\ntemp 0–1.5"]
        MM2508["mistral-medium-2508 (existing)\n128K ctx | $0.40/$2.00"]
        CL["codestral-latest 🔧\n256K ctx (was 128K)"]
        C2508["codestral-2508 🔧\n256K ctx (was 128K)"]
    end

    DVP -->|"newer reasoning model"| DC
    DVF -->|"newer chat model"| DC
    MM2604 -->|"successor to"| MM2508

    classDef new fill:#d4edda,stroke:#28a745
    classDef fixed fill:#fff3cd,stroke:#ffc107
    classDef existing fill:#f8f9fa,stroke:#6c757d
    class DVP,DVF,MM2604 new
    class CL,C2508 fixed
    class DC,DR,MM2508 existing

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    subgraph DeepSeek["DeepSeek Provider"]
        DVP["deepseek-v4-pro ✨\n1M ctx | $0.435/$0.87\ncached $0.003625\nNo temperature cap"]
        DVF["deepseek-v4-flash ✨\n1M ctx | $0.14/$0.28\ncached $0.0028\ntemp 0–2"]
        DC["deepseek-chat (existing)\n1M ctx | $0.14/$0.28"]
        DR["deepseek-reasoner (existing)\n1M ctx | $0.14/$0.28"]
    end

    subgraph Mistral["Mistral Provider"]
        MM2604["mistral-medium-2604 ✨\n256K ctx | $1.50/$7.50\ntemp 0–1.5"]
        MM2508["mistral-medium-2508 (existing)\n128K ctx | $0.40/$2.00"]
        CL["codestral-latest 🔧\n256K ctx (was 128K)"]
        C2508["codestral-2508 🔧\n256K ctx (was 128K)"]
    end

    DVP -->|"newer reasoning model"| DC
    DVF -->|"newer chat model"| DC
    MM2604 -->|"successor to"| MM2508

    classDef new fill:#d4edda,stroke:#28a745
    classDef fixed fill:#fff3cd,stroke:#ffc107
    classDef existing fill:#f8f9fa,stroke:#6c757d
    class DVP,DVF,MM2604 new
    class CL,C2508 fixed
    class DC,DR,MM2508 existing

Loading

_{Reviews (1): Last reviewed commit: "improvement(models): add DeepSeek V4 + M..." | Re-trigger Greptile}