Ollama

What's Changed

• app: fix ChatGPT model selector spacing
• mlxrunner: Evict prefix cache snapshots from the active conversation
• mlxrunner: check system free memory and wait for evicted runners before loading the next MLX model
• llm: raise token repeat limit to 100 and return error instead of incomplete result
• mlx: scope array lifetimes instead of pinning and sweeping
• llm: keep gemma3n projector off the CPU
• app: refresh Apps layout and command copy feedback
• MLX and llama.cpp updates

Full Changelog: https://github.com/ollama/ollama/compare/v0.34.0...v0.34.1-rc1

Use Ollama models in ChatGPT Desktop

Ollama models can now be used directly in ChatGPT Desktop, so you can keep your existing workflow while running open models. Setup is available from the Ollama app on MacOS.

This release also improves structured output performance on Apple Silicon, adds support for OpenAI-compatible client tool search and response compaction.

Full Changelog: https://github.com/ollama/ollama/compare/v0.33.3...v0.34.0

What's Changed

• gemma4 now supports images and audio on MLX engine
• Report cached prompt tokens
• Honor GGUF model defined default parameters
• MLX, MLX-C, llama.cpp update

New Contributors

• @marcelpetrick made their first contribution in https://github.com/ollama/ollama/pull/17579

Full Changelog: https://github.com/ollama/ollama/compare/v0.33.2...v0.33.3

What's Changed

• Ollama's app now follows the system appearance again, restoring dark mode support
• Fixed the macOS app to properly hand off to an already-running instance instead of starting a second one
• The Claude Desktop proxy no longer interrupts in-flight requests when the model catalog updates

Full Changelog: https://github.com/ollama/ollama/compare/v0.33.1...v0.33.2

What's Changed

• MLX: Qwen3.8 Flash Next support
• cmake: make external compat patches idempotent
• MLX and llama.cpp update
• mlxrunner: add structured output support
• mlxrunner: avoid Metal GPU timeouts when loading models from slow storage

New Contributors

• @pd95 made their first contribution in https://github.com/ollama/ollama/pull/17948

Full Changelog: https://github.com/ollama/ollama/compare/v0.33.0...v0.33.1

What's Changed

Claude Desktop

Developers can now easily configure Claude Desktop to seamlessly work with Ollama as a third-party gateway provider.

Improved caching

• Fixed a hang where agent clients that cancel long prefills
• Prefill restore points are now trustworthy by construction: a cancelled prefill keeps every restore point it crossed, so retries resume where they stopped instead of restarting from scratch
• Resumed prefills no longer record restore points that fail to cover what they claim; on models with recurrent layers this previously forced a request matching 46k of 47k tokens to reprocess from zero
• Disabled Claude Code's "tokens left" token-countdown system message, which Ollama moved to the front of the prompt and broke the KV cache on every request

Other improvements

• DeepSeek Harness launcher now falls back to npx when the global npm install fails, with Windows command-shim support
• MLX dependency update (#17886)
• Fixed broken default packaging caused by macOS-specific assumptions affecting Linux/Windows builds

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.15...v0.33.0

What's Changed

• New desktop onboarding flow on first launch
• Caches resolved model metadata between requests, cutting time-to-first-token by roughly half (TTFT dropped from ~995 ms to ~524 ms in benchmarks)
• Fixes a bug where chat and generate could wedge after a mid-stream parser error
• Qwen 3.8 system messages are now normalized so non-leading system messages are handled consistently
• MLX and llama.cpp dependency updates

New Contributors

• @gaugarg-nv made their first contribution in https://github.com/ollama/ollama/pull/17752

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.14...v0.32.15

What's Changed

• llm: transcode WebP images for llama-server
• renderers/qwen: tolerate non-leading system messages

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.13...v0.32.14

What's Changed

• qwen3.8: support developer instructions

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.12...v0.32.13

Qwen 3.8 27B

This release adds the support of Qwen 3.8 27B. Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks.

ollama run qwen3.8:27b

For Apple Silicon devices, Ollama has in particular optimized for maximum performance and output quality suitable for repeated tasks and coding agents.

ollama run qwen3.8:27b-mlx

What's Changed

• ollama launch dsh now supports DeepSeek Harness, DeepSeek's open-source agent harness
• ollama launch muse now supports Muse Code, Meta's agentic coding CLI
• The OpenAI-compatible Responses API now supports web search
• Muse Glimmer template updates

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.10...v0.32.11

What's Changed

• Models that don't set a repeat_penalty now default to 1.0 (off) instead of 1.1, matching other engines and speeding up speculative decoding; set a per-model parameter if an older model repeats itself.
• Faster prefill on NVFP4 MLX models with a global scale, about 7–8% on Qwen3.6 and Muse Glimmer.
• Fixed blob verification being skipped when an OCI manifest's config and layer share a digest.

New Contributors

• @vigneshakaviki made their first contribution in https://github.com/ollama/ollama/pull/15504

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.8...v0.32.10-rc1

NVIDIA Nemotron 3.5 Lightning

NVIDIA Nemotron 3.5 Lightning is an open 30B mixture-of-experts (MoE) model with 3B active parameters built for that execution layer of always-on agents. It is designed for harnesses like OpenClaw and Hermes Agent – all supported by the NVIDIA NemoClaw open source security and management stack for running always-on AI agents.

ollama run nemotron-3.5-lightning

What's Changed

• Added the Nemotron 3 architecture
• Handle boundary condition in Muse Glimmer function calling parser

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.8...v0.32.9

Muse Glimmer

Muse Glimmer is now available on all platforms. Muse Glimmer can power coding agent applications such as Claude Code, Codex, Pi and more, as well as long-running personal assistants such as OpenClaw and Hermes.

Ollama's MLX engine provides state-of-the-art performance on Apple Silicon for this model, with support for DFlash and image input as of Ollama 0.32.7.

To download and run Muse Glimmer locally:

ollama run muse-glimmer

To run Muse Glimmer with Claude Code, download Ollama and run:

ollama launch claude --model muse-glimmer

For a lighter-weight coding agent, try Pi:

ollama launch pi --model muse-glimmer

For personal assistant frameworks such as OpenClaw and Hermes, use:

ollama launch openclaw --model muse-glimmer

ollama launch hermes --model muse-glimmer

What's Changed

• Add Muse Glimmer support for NVIDIA, AMD, and additional platforms

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.7...v0.32.8

Muse Glimmer

Note: Muse Glimmer is currently available via initial support via Ollama's MLX engine on Apple Silicon. Additional support and optimizations for Apple Silicon, NVIDIA, AMD, and other platforms will be available in the coming days.

Muse Glimmer, Meta's newest open model and the first released by Meta Superintelligence Labs, is now available on Ollama. It's a 30B multimodal model purpose-built for agent workloads that run locally.

With Ollama, you can now use Muse Glimmer to power coding agent applications such as Claude Code, Codex, Pi and more, as well as long-running personal assistants such as OpenClaw and Hermes.

Ollama's MLX engine provides state-of-the-art performance on Apple Silicon for this model, with support for DFlash and image input as of Ollama 0.32.7.

To download and run Muse Glimmer locally:

ollama run muse-glimmer:30b-mlx

To run Muse Glimmer on Apple Silicon with Claude Code, download Ollama and run:

ollama launch claude --model muse-glimmer:30b-mlx

For a lighter-weight coding agent, try Pi:

ollama launch pi --model muse-glimmer:30b-mlx

For personal assistant frameworks such as OpenClaw and Hermes, use:

ollama launch openclaw --model muse-glimmer:30b-mlx

ollama launch hermes --model muse-glimmer:30b-mlx

What's Changed

• Qwen3.5 is faster on Apple GPUs: the MLX engine now uses the model's MTP head for speculative decoding automatically
• /v1/chat/completions streaming now matches OpenAI's wire format: role only on the first chunk, finish_reason on its own chunk, and usage in a separate chunk with stream_options.include_usage.
• Truncated OpenAI responses now report finish_reason: "length" instead of "tool_calls".
• ollama run kimi-k3 now offers kimi-k3:cloud for cloud-only models that publish no default tag, instead of failing.
• TUI fixes: pipe-delimited prose no longer renders as a table, Enter accepts the highlighted @ file completion, and /prompt scrolling is no longer laggy.
• Experimental image generation has been temporarily removed. Continue using 0.32.5 for image generation support
• Updated the MLX and llama.cpp engines.

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.5...v0.32.6-rc0

What's Changed

• Fixed an MLX Metal bug that could reduce output quality for NVFP4 models, particularly Laguna.

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.4...v0.32.5

What's Changed

• Support Laguna on Apple GPUs via the MLX engine
• Quantize draft-model output heads at the requested type when creating speculative-decoding drafts.
• Fixed Qwen3 MoE decoding for differently-quantized experts, plus faster packed gate/up projection (~4–9% on M5 Max).

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.3...v0.32.4

What's Changed

• Fixed model downloads that stall before sending data.
• Improved integrations: restored Claude Code Channels, fixed Anthropic thinking streams, and made Hermes Desktop respect --force-build.
• Expanded GPU support with CUDA on Windows ARM64, B200 support through CUDA 12, and lower memory use on Linux CUDA/ROCm iGPUs.
• Added chat, thinking, and tool calling support for Laguna 2.1 models, including a Metal inference fix.
• Fixed GLM tool calls being silently dropped at the end of generation.
• Updated the MLX and llama.cpp engines.

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.1...v0.32.3

What's Changed

• Improved Gemma 4 tool calling and multi-turn reasoning, including more reliable tool-response continuations
• Fixed a recurrent MLX model cache leak that could increase memory use across requests, and improved cache snapshot performance
• MLX text model loading now respects OLLAMA_LOAD_TIMEOUT
• Agent web search and fetch now tell users to run ollama signin when authentication is required
• The interactive agent now receives the current working directory for better project context
• Fixed ollama launch so choosing Pick another model for a deprecated model passed with --model opens the model picker
• Updated VS Code setup documentation for the official Ollama extension

Full Changelog: https://github.com/ollama/ollama/compare/v0.32.0...v0.32.1-rc0

What's Changed

• New interactive agent experience: running ollama now launches an agent to help you code and delegate work

❯ ollama
Ollama 0.32.0

▸ Chat, Code, & Work (glm-5.2:cloud)
    Chat with models, code, search the web, and delegate real work

• Renamed the Codex App integration to ChatGPT: use ollama launch chatgpt (and --restore to return to your usual ChatGPT profile)
• Simplified integration selection: the ollama launch menu now only offers the most popular integrations (other integrations can be accessed through ollama launch
• Warns before launching older agent models: CodeLlama, Qwen2.5(-coder), Llama 3.x, Mistral, StarCoder, and the base DeepSeek-R1 tags now prompt a deprecation warning before ollama launch continues

Full Changelog: https://github.com/ollama/ollama/compare/v0.31.2...v0.32.0

What's Changed

• Enabled flash attention on older NVIDIA GPUs (compute capability 6.x)
• iGPU can now offload vision models with padding to fit available memory
• Fixed structured output for thinking models when thinking is disabled
• Hardened GGUF model creation
• ollama launch for Claude Code now disables telemetry by default
• Fixed loading models on paths with non-UTF-8 characters
• Updated the MLX and llama.cpp engines

New Contributors

• @kevinpark1217 made their first contribution in https://github.com/ollama/ollama/pull/16949

Full Changelog: https://github.com/ollama/ollama/compare/v0.31.1...v0.31.2

Faster Gemma 4 on Apple Silicon

Gemma 4 is now significantly faster in Ollama on Apple Silicon, generating tokens nearly 90% faster on average across a coding-agent benchmark by leveraging multi-token prediction (MTP). Ollama auto-tunes how many tokens to draft as it runs, so the speedup is on by default, requires no configuration, and does not change the model's output.

What's Changed

• Tightened Gemma 4 MoE model loading in the MLX engine
• Updated the MLX engine to the latest version, including a new small-batch matmul kernel
• Updated the underlying llama.cpp engine to build 9840
• Improved Gemma 4 multi-token prediction (MTP) performance

Full Changelog: https://github.com/ollama/ollama/compare/v0.30.12...v0.31.1

What's Changed

• tools: ignore braces inside JSON strings when detecting tool call end by @aditya-786 in https://github.com/ollama/ollama/pull/16937
• mlx: bump dependency by @dhiltgen in https://github.com/ollama/ollama/pull/16935
• llama.cpp update by @dhiltgen in https://github.com/ollama/ollama/pull/16960

New Contributors

• @aditya-786 made their first contribution in https://github.com/ollama/ollama/pull/16937

Full Changelog: https://github.com/ollama/ollama/compare/v0.30.11...v0.30.12-rc0

What's Changed

• launch: add thinking capability detection to opencode by @hoyyeva in https://github.com/ollama/ollama/pull/15434
• launch: auto-install Claude Code by @hoyyeva in https://github.com/ollama/ollama/pull/16802
• launch: auto-install opencode when missing by @hoyyeva in https://github.com/ollama/ollama/pull/16806
• discover: fix inverted iGPU/dGPU Vulkan classification on Windows hybrid graphics by @Sahil170595 in https://github.com/ollama/ollama/pull/16669
• mlxrunner: unify and tune speculative decoding by @jessegross in https://github.com/ollama/ollama/pull/16791
• launch/codex: detect model drift when Codex App UI switches by @BruceMacD in https://github.com/ollama/ollama/pull/16864
• llama: add sm_86 architecture to cuda_v13_windows preset by @anishesg in https://github.com/ollama/ollama/pull/16834
• llm: size mmproj offload by projector memory by @dhiltgen in https://github.com/ollama/ollama/pull/16866
• docs: document max think level by @ParthSareen in https://github.com/ollama/ollama/pull/16877
• llm: preserve generation headroom for shifted prompts by @ParthSareen in https://github.com/ollama/ollama/pull/16856
• llama: default qwen2.5vl window attention metadata by @dhiltgen in https://github.com/ollama/ollama/pull/16868
• llm: use host Vulkan loader on Windows by @dhiltgen in https://github.com/ollama/ollama/pull/16869
• mlx: update and fix CUDA JIT packaging by @dhiltgen in https://github.com/ollama/ollama/pull/16871
• llm: fix ollama ps double-counting mmap'd weights on partial offload by @discobot in https://github.com/ollama/ollama/pull/16709
• docs: redesign docs landing and integrations overview by @hoyyeva in https://github.com/ollama/ollama/pull/16807
• server: align generate with native chat templates by @dhiltgen in https://github.com/ollama/ollama/pull/16878
• jetson: add CC 87 for CUDA v13 by @dhiltgen in https://github.com/ollama/ollama/pull/16628
• llama.cpp version update by @dhiltgen in https://github.com/ollama/ollama/pull/16548

New Contributors

• @Sahil170595 made their first contribution in https://github.com/ollama/ollama/pull/16669
• @anishesg made their first contribution in https://github.com/ollama/ollama/pull/16834
• @discobot made their first contribution in https://github.com/ollama/ollama/pull/16709

Full Changelog: https://github.com/ollama/ollama/compare/v0.30.10...v0.30.11-rc0

What's Changed

• Command A and North family models now run on Apple Silicon with the MLX engine
• Updated the underlying llama.cpp engine to build 9672
• Fixed build artifacts for MLX

Full Changelog: https://github.com/ollama/ollama/compare/v0.30.9...v0.30.10

What's Changed

• Support for Cohere2Moe architecture
• Fixed LFM2 parser/render for cases where thinking was not emitted
• Fixed issue where ollama launch claude and other coding agent or assistant use cases would only output one token
• Ollama will now return an error if a single message is larger than the current context window

Full Changelog: https://github.com/ollama/ollama/compare/v0.30.8...v0.30.9-rc1