Models/Mellum2.1 Thinking

Coding / JetBrains

Mellum2.1 Thinking

A 12B-total, 2.5B-active open-weight thinking model for coding and tool use. Start with JetBrains’ official GGUF on a suitable desktop, or serve the BF16 checkpoint with vLLM on a larger GPU.

Verified sourceReleased Sep 20, 2026Source checked 10/14/2026

LOCALRENTED GPUOPEN WEIGHTS

Freshness 91 out of 100. Release age: 25 days old, Sep 20, 2026. Measures age, not quality.

Popularity 47 out of 100. 2,445 downloads in 30 days; 143 lifetime likes. Checked Oct 15, 2026. Measures adoption, not quality.

What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

A new release starts at 100. The score halves every 180 days: 100 × 2^(−age in days / 180). When the release date is unavailable, a repository date is a fallback—not proof of when the model was released. Missing dates stay unscored.

Check the date source

Popularity · adoption

Downloads over the past 30 days carry 80% of the score; lifetime likes carry 20%. Both use a logarithmic scale, capped at 10 million downloads and 100,000 likes. If likes are unavailable, downloads carry all the weight. Missing download statistics stay unscored.

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source

Scores as of Oct 15, 2026. Statistics checked Oct 15, 2026. Statistics are marked for refresh after 72 hours.Full scoring method

WORKFLOWS FOR THIS MODEL 1

What do you want to do?

Choose a task. Get the software, files, settings, and steps to run it.

MAKE IT PERSONALChoose your hardware

Coding

Code locally with Mellum2.1 Thinking

Serve Mellum2.1 Thinking · 12B BF16 privately, connect VS Code and Cline, then review and test a real code change.

Freshness 91 out of 100. Release age: 25 days old, Sep 20, 2026. Measures age, not quality.

VS Code + ClineOllama

2 packagesmacOS / Windows / Linux

Open coding guide

Open a guide for its software, hardware requirements, and setup steps.

At a glance

Parameters: 12B total · 2.5B active
Architecture: Mixture of experts
Context length: 131,072 architectural; usable context depends on memory
License: Apache 2.0
Software: Ollama, llama.cpp, VS Code + Cline, vLLM

MY HARDWARE

Will it run on your machine?

Save your machine to see a personalized rating and its reasoning.

Add my hardware

Best for

Private coding assistance in an editor with a self-hosted endpoint.
Repository tasks that use tools and inspect file changes.

Tradeoffs

The Q4 file needs additional runtime and context memory; validate Cline tool use on your exact package.
The BF16 checkpoint needs substantially more GPU memory than its 2.5B active parameter count suggests.

Ways to run it

Ollama or llama.cpp · official Q4_K_M GGUF · Windows, macOS, or Linux; confirm tool support and context on your exact machine

JetBrains publishes an 8.1 GB Q4_K_M package. The file size alone is not a complete memory or coding-agent guarantee.

ollama run hf.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking-GGUF:Q4_K_M

Official instructions

vLLM · BF16 · Linux GPU server; Windows and macOS can connect as clients

Full-precision server route; account for roughly 24 GB weights plus KV cache and runtime memory.

vllm serve JetBrains/Mellum2.1-12B-A2.5B-Thinking --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser hermes

Official instructions

Explore its uses

CODINGAGENTSREASONINGTOOL CALLINGOLLAMALLAMA.CPPVLLM

REFERENCE LIBRARY

Sources and files

Primary sources and files behind this model page.

2 SOURCES

01JetBrains model card, vLLM recipe and licensehuggingface.co 02JetBrains official GGUF packages and Ollama commandshuggingface.co

Keep exploring

Coding

Qwen

Qwen2.5 Coder · 7B

Code generation, repair, and reasoning in a compact open-weight coding model.

Freshness 5 out of 100. Repository age: 758 days old, Sep 17, 2024. Measures age, not quality.

Mistral Small 4 · 119B

Mistral Small 4 119B A6B is an Apache-2.0 licensed hybrid model from mistralai that combines general instruction, reasoning, and Devstral-style capabilities in one model. It uses a Mixture-of-Experts architecture with 119B total parameters, 6.5B activated per token, 128 experts with 4 active, and a 256k context length. It supports multimodal input (text and image) with text output, multilingual support across dozens of languages, system prompts, function calling, JSON output, and configurable reasoning effort. The model card also mentions optional efficiency resources: an eagle head for speculative decoding and an NVFP4 checkpoint. It is tagged for vLLM and offered in safetensors format.

Freshness 36 out of 100. Repository age: 265 days old, Jan 23, 2026. Measures age, not quality.

MODEL SIZE119B total · 6.5B active

LOCALRENTED GPU

Repository opened Jan 23, 2026

Coding

Qwen

Qwen3-Coder-Next · 80B A3B

Qwen3-Coder-Next is an open-weight text-generation model from Qwen/Qwen3-Coder-Next, licensed under apache-2.0. It is designed specifically for coding agents and local development. The model has 80B total parameters with 3B activated parameters, a 262,144-token native context length, and a hybrid architecture combining Gated DeltaNet, Gated Attention, and Mixture-of-Experts layers. It supports only non-thinking mode and does not generate thinking blocks.

Freshness 37 out of 100. Repository age: 258 days old, Jan 30, 2026. Measures age, not quality.

MODEL SIZE80B total / 3B active

LOCALRENTED GPU

Repository opened Jan 30, 2026