Mellum2-12B-A2.5B-Thinking

Mellum2-12B-A2.5B-Thinking is a reasoning model from JetBrains in the Mellum2 12B-A2.5B family. 6 benchmarks count toward its score, in 5 categories.

partialShows if the model has enough results for an index.
IndexOverall score out of 100.Unranked
CoverageShare of the index weight with results.65%
SpeedOutput tokens per second.
Input / 1MUS dollars per 1M input tokens.N/A
Output / 1MUS dollars per 1M output tokens.N/A
ContextMaximum tokens in one request.128K
EloLMArena rating and rank.N/A

The index is a score out of 100. The ± range shows how much it can change.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
29.9
CodingCode writing and repair.
40.6
ReasoningLogic problems and puzzles.
N/A
MultimodalTasks with images and text.
N/A
KnowledgeFacts and expert knowledge.
35.8
MultilingualTasks in many languages.
38.8
InstructionTasks with strict rules in the prompt.
16.1
MathMath problems.
N/A

Results

6 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result as a score out of 100.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
MMLU-ReduxKnowledge86.2%40.1Qwen
Instruction-Following EvalInstruction76.5%16.1Jeffrey Zhou et al.
LiveCodeBench v6Coding69.9%40.6LiveCodeBench maintainers
Graduate-Level Google-Proof Q&AKnowledge57.6%31.4David Rein et al.
GPQA DiamondKnowledge57.6%31.4David Rein et al.
Berkeley Function Calling Leaderboard v4Agentic45.6%29.9Arcee AI
EuroEval DutchMultilingual31.2%41.2EuroEval
EuroEval GermanMultilingual27.3%36.4EuroEval

6 benchmarks count, from 8 of 8 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsEuroEval, collected directlyMIT — the leaderboard site and its CSV routes are in the licensed repository

More from JetBrains

Mellum2-12B-A2.5B-InstructUnranked