Skip to content
TrustList
M1
AI Model

Mellum2.1

New· 4

JetBrains' open 12B mixture-of-experts coding model with 2.5B parameters active per token under Apache 2.0

About Mellum2.1

JetBrains published Mellum2.1 in October 2026, around 8 October, as an update of its Mellum2 coding model. The weights are open and carry the Apache 2.0 license.

Design and size

The model is a mixture-of-experts system with 12 billion parameters in total and 2.5 billion active for each token. Its model card gives a context window of 131,072 tokens.

Agentic coding and hard problems

JetBrains aims it at agentic coding work: operating inside a repository, running commands and calling tools. It is also meant for hard coding, math and reasoning problems where no agent setup is involved.

Faster serving

A technique called multi-token prediction speeds up a single request by roughly 1.6 times, according to the announcement. It also says that on one H200 GPU under heavy load the model pushes through nearly double the tokens that Qwen3.5-9B manages. Charts in the post set Mellum2.1 against Mellum2, Qwen3.5-9B and Gemma 4 E4B, naming LiveCodeBench, AIME, GPQA, BFCL, IFEval and SWE-Verified as the tests. These are the vendor's own runs, and the text gives no numbers.

Getting and running it

Hugging Face carries the weights. GGUF builds are offered for llama.cpp, Ollama and LM Studio, and support through vLLM is listed as coming soon. No hosted inference provider serves it yet and there is no hosted API, so teams would run it on their own hardware. That profile suits groups that prefer to keep source code on machines they control, since the weights can be run locally and no outside service is needed.

TrustList summary

Mellum2.1 updates JetBrains' Mellum2 coding model. It is published with open weights, licensed under Apache 2.0. The model uses a mixture-of-experts layout: 12 billion parameters in total, 2.5 billion active, and JetBrains dates its release to around 8 October 2026.

What Mellum2.1 does

  • Agentic coding: works inside a repository, runs commands and calls tools.
  • Coding and reasoning: tackles hard coding, maths and reasoning problems without an agent setup.
  • Speed: JetBrains says multi-token prediction makes one request roughly 1.6 times quicker.
  • Throughput: on one H200 GPU under heavy load, serves almost twice as many tokens as Qwen3.5-9B, per JetBrains' announcement.

Who it works with

Its open weights under Apache 2.0 allow teams to host the model themselves.

Key facts

  • Developer: JetBrains
  • Context window: 131,072 tokens, per the model card
  • Licence: Apache 2.0

Summary written by TrustList from the business's own published information, Nov 2026.

Benchmarks & AI stats

BenchmarkOfficialCommunity avg
No benchmark scores yet. Be the first to add one.

“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).

Sign in to add a benchmark score for this model.

Trust Score

4/ 100
Trust Score: New

An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.

Verification
0/100 · 20%
Reviews
0/100 · 30%
Awards
0/100 · 15%
Transparency
18/100 · 20%
Engagement
0/100 · 15%
Joining soon
Recommendations
Coming soon
Complaints
Coming soon

Updated 10/9/2026

Request a demo or quote from Mellum2.1

Protected by reCAPTCHA - Google Privacy Policy and Terms apply.

By sending, you agree we may share your request and contact details with the provider once you confirm your email.

Reviews

Write the first review of Mellum2.1

Used it? Your experience helps other buyers decide.

Write a review

Questions & answers

No questions yet. Be the first to ask about Mellum2.1.

Mellum2.1 alternatives

Similar listings buyers compare against this one.

Open-weight translation model from Bilibili's Index team, covering 150 languages and following glossary and style instructions

Cantina Security's first open-weight security model, a post-trained GLM-5.3-Flash for vulnerability research and exploit checks

Black Forest Labs' image generation and editing model with bounding-box layouts, multiple reference images and 4K output

Low-latency version of ElevenLabs' Eleven v4 speech model, aimed at phone and chat voice agents in 90+ languages

Anthropic's smallest and fastest Claude 5.5 model, available on its own API and three cloud platforms

Ai2's open-weight 8B model that writes cited scientific reports from a research question and retrieved literature

Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems

Open-source 2B decision model from AWS's Strands Agents project that picks options and returns scores with a reliability figure