LFM2.5 8B A1B Hermes Abliterated

Not independently scored by Artificial Analysis — its base model DuoNeural/LFM2.5-8B-A1B-Abliterated is the closest proxy

DuoNeural's fine-tune of an abliterated LFM2.5-8B-A1B, Liquid AI's mixture of experts with about 1B active parameters, sold as an agentic coder. Abliterated means the refusal behaviour was removed on purpose: the model does not decline requests. Licence: Liquid's own community licence, not an open-source one. It is here because we measured it, not as a recommendation.

BigCodeBench-Hard pass@1 5% (7/148), via official protocol · llama.cpp Q4_K_M · one request at a time · a shared cluster GPU · mean of 3 runs, range 0.0 pp. The brutal counterpart to HumanEval — where HumanEval saturates near the top, BCB-Hard spreads the field, so this is the number that actually separates coding ability. does not fit the token budget; repeats itself until it runs out; 6 of 148 answers hit the token limit (effective ceiling 96%)

84.5% ± 2.4 over 1390 problems

Q4_K_M · a shared cluster GPU · KV f16 · 1 run. This is the non-agentic half of BFCL: does the model call the right function with the right arguments. It says nothing about how it behaves over a long agentic conversation, which is a separate measurement. The score is BFCL’s own: the unweighted mean of four groups (simple, multiple, parallel, parallel multiple), where simple is itself the mean of Python, Java and JavaScript. Every group counts the same whatever its size, and irrelevance detection — which we also measure, another 240 problems — is not part of it. The margin follows that same weighting.

All 13 RULER tasks, one request at a time, KV cache f16, bench commit c3f5e3b4, seed 42. The score is the mean of the 13; the margin is two standard deviations of the sampling, so it grows as the sample shrinks and as the score drops.

This is not a coding score and must not be read as one. RULER asks whether the model finds and follows a fact buried in a long text. Quantising hits the two differently, so a model that still retrieves at length may already have lost the code.

Weak at code and fine at tool calls: 4.7% on BigCodeBench over three runs, 84.5% on BFCL non_live in one run. Use it to route tools, not to write code.