LFM2.5-8B-A1B vs Qwen3.5 4B
Wins 1 of 6 areas
Following instructions
Wins 4 of 6 areas
Coding · Reasoning · Math · Long documents
Qwen3.5 4B is the stronger all-rounder.LFM2.5-8B-A1B is better at following instructions.
Scores updated · 22 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking02Qwen3.5 4B2 of 3 tests · 1 tie
- CodingWriting and fixing software02Qwen3.5 4B2 of 2 tests
- MathCompetition and research-level math01Qwen3.5 4B1 of 1 test
- Long documentsFinding answers in very long texts01Qwen3.5 4B1 of 1 test
- Following instructionsDoing exactly what it is asked10LFM2.5-8B-A1B1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Math, long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Qwen3.5 4B pulls ahead
- US invitational high-school math exam problemsAIME 2026+39.7points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+25.8points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+13.7points ahead
Where LFM2.5-8B-A1B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+39.7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+3.6points ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.5 4B
- Terminal-Bench HardQwen3.5 4B by 13.74.518.2+13.7
- SciCodeQwen3.5 4B by 8.37.816.1+8.3
ReasoningQwen3.5 4B
- GPQA DiamondQwen3.5 4B by 25.851.377.1+25.8
- Humanity's Last ExamQwen3.5 4B by 36.99.9+3
- CritPttie00tie
FactsEven
- AA-Omniscience · Non-hallucinationLFM2.5-8B-A1B by 39.753.113.4+39.7
- AA-Omniscience · AccuracyQwen3.5 4B by 5.89.315.1+5.8
Following instructionsLFM2.5-8B-A1B
- IFBenchLFM2.5-8B-A1B by 3.655.652+3.6
Other results12 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- τ²-Bench Telecom (AA run)Qwen3.5 4B by 7616.192.1+76
- MATH500 (Pass@1)Qwen3.5 4B by 72.58.380.8+72.5
- τ²-Bench (Retail)Qwen3.5 4B by 32.139.871.9+32.1
- AA Agentic IndexQwen3.5 4B by 27.15.432.5+27.1
- AA-OmniscienceLFM2.5-8B-A1B by 25.1-33.2-58.4+25.1
- Artificial Analysis Coding IndexQwen3.5 4B by 175.622.6+17
- AIME25 no toolsQwen3.5 4B by 6.842.549.3+6.8
- BFCL v3Qwen3.5 4B by 6.364.871.1+6.3
- AA IntelligenceQwen3.5 4B by 5.97.213.1+5.9
- IFEvalLFM2.5-8B-A1B by 5.691.886.2+5.6
- BFCLv4tie49.750.6tie
- Tau² Telecomtie88.187.7tie
Questions people ask
Which is better, LFM2.5-8B-A1B or Qwen3.5 4B?
Qwen3.5 4B wins four of the six areas where both have results: coding, reasoning, math and long documents. LFM2.5-8B-A1B wins following instructions. They are level on facts.
Which is better for coding?
Qwen3.5 4B. It wins 2 of the 2 coding tests both models report; LFM2.5-8B-A1B wins none.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 10 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 12 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.