Ollama v0.40.14 — Release Notes & Engineering Impact

Source available — automated, not human-reviewed

Component: ollama · Channel: stable. Automated analysis, not a compatibility certification. How this works.

Analysis updated: 2026-11-03.

Compatibility not verified — check source notes

📄 .MD

⚡ Improvements 6 🔧 Bug Fixes 1

⚡ Executive Summary (TL;DR)

Ollama v0.40.14 aligns OpenAI-compatible endpoint sampler defaults with packaged model settings and speeds up MLX top-k sampling for large vocabularies. It also improves connection resiliency for ChatGPT routing, fixes web search response decoding on `/api/chat`, and applies appropriate output token limits for Claude Code.

Visit Ollama Official Site ↗ Official Release Source ↗ All Ollama Releases 📡 Ollama RSS

Who Does This Impact? (Role-Based Analysis)

💻 Software Engineers & Devs

Engineers using OpenAI-compatible endpoints should note that omitting `temperature` or `top_p` now falls back to model-packaged defaults rather than hardcoding to 1.0.

📊 Product & Engineering Leads

Integrations relying on cloud web search, Claude Code limits, and ChatGPT routing will experience improved reliability and performance.

⚙️ DevOps & Security

No deployment or hosting changes are required, though updates resolve transient ChatGPT connection failures and routing errors in CLI launch commands.

⚡ Performance & Enhancements (6)

`ollama launch claude` applies model output token limits to prevent running into the 128k default cap
OpenAI-compatible endpoints default to model-packaged sampler settings (`temperature`/`top_p`) when parameters are omitted by clients
`/v1/responses` preserves readable input content, including prior assistant refusals
Ollama desktop application retries ChatGPT disconnect failures automatically
`ollama launch codex` verifies ChatGPT routing before configuration and reports explicit errors when routing is unavailable
Faster top-k sampling performance on MLX when handling large vocabularies

🔧 Bug Fixes & Patches (1)

Cloud web search responses on `/api/chat` (including compressed payloads) are now decoded correctly

📄 View Raw Release Notes (Original Developer Log)

## What's Changed * `ollama launch claude` now applies the model's output token limit for Claude Code, so cloud models no longer run into the 128k default cap * Cloud web search responses on `/api/chat` are now decoded correctly, including compressed responses * OpenAI-compatible endpoints no longer override a model's packaged sampler defaults — clients that send no `temperature`/`top_p` get the model's own values instead of 1.0, matching `/api/chat` * `/v1/responses` now preserves readable input content, including prior assistant refusals, instead of flattening it * Ollama's app can now retry ChatGPT disconnect failures instead of leaving the connection in a broken state * `ollama launch codex` verifies ChatGPT routing before configuring, and reports a clear error when routing is unavailable * Faster top-k sampling on MLX for large vocabularies **Full Changelog**: https://github.com/ollama/ollama/compare/v0.40.13...v0.40.14

← Older Version Ollama v0.40.13 Newer Version → Ollama v0.40.14-rc0