LFM2.5-8B-A1B vs Qwen3 30B A3B 2507 Thinking
Wins 1 of 7 areas
Following instructions
Wins 4 of 7 areas
Coding · Reasoning · Math · Long documents
Qwen3 30B A3B 2507 Thinking wins more areas, narrowly.LFM2.5-8B-A1B is better at following instructions.
Scores updated · 21 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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 30B A3B 2507 Thinking2 of 3 tests · 1 tie
- CodingWriting and fixing software01Qwen3 30B A3B 2507 Thinking1 of 2 tests · 1 tie
- MathCompetition and research-level math01Qwen3 30B A3B 2507 Thinking1 of 1 test
- Long documentsFinding answers in very long texts01Qwen3 30B A3B 2507 Thinking1 of 1 test
- Following instructionsDoing exactly what it is asked10LFM2.5-8B-A1B1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
- FactsGetting facts right instead of making them up11Even1 each
Agents, 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 30B A3B 2507 Thinking pulls ahead
- US invitational high-school math exam problemsAIME 2026+38.3points ahead
- Code for real scientific research problemsSciCode+25.2points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+19.4points ahead
Where LFM2.5-8B-A1B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+39.6points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+4.9points ahead
Every test, side by side
All 21 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3 30B A3B 2507 Thinking
- SciCodeQwen3 30B A3B 2507 Thinking by 25.27.833+25.2
- Terminal-Bench Hardtie4.55.3tie
ReasoningQwen3 30B A3B 2507 Thinking
- GPQA DiamondQwen3 30B A3B 2507 Thinking by 19.451.370.7+19.4
- Humanity's Last ExamQwen3 30B A3B 2507 Thinking by 3.46.910.3+3.4
- CritPttie00.3tie
FactsEven
- AA-Omniscience · Non-hallucinationLFM2.5-8B-A1B by 39.653.113.5+39.6
- AA-Omniscience · AccuracyQwen3 30B A3B 2507 Thinking by 79.316.3+7
MathQwen3 30B A3B 2507 Thinking
- AIME 2026Qwen3 30B A3B 2507 Thinking by 38.35088.3+38.3
Long documentsQwen3 30B A3B 2507 Thinking
- AA-LCRQwen3 30B A3B 2507 Thinking by 61.3061.3+61.3
Following instructionsLFM2.5-8B-A1B
- IFBenchLFM2.5-8B-A1B by 4.955.650.7+4.9
Other results10 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.
- MATH500 (Pass@1)Qwen3 30B A3B 2507 Thinking by 78.28.386.5+78.2
- Tau² TelecomLFM2.5-8B-A1B by 66.288.121.9+66.2
- AIME25 no toolsQwen3 30B A3B 2507 Thinking by 29.242.571.7+29.2
- AA-OmniscienceLFM2.5-8B-A1B by 22.9-33.2-56.1+22.9
- τ²-Bench (Retail)Qwen3 30B A3B 2507 Thinking by 16.339.856.1+16.3
- τ²-Bench Telecom (AA run)Qwen3 30B A3B 2507 Thinking by 1216.128.1+12
- BFCL v3Qwen3 30B A3B 2507 Thinking by 8.664.873.4+8.6
- AA IntelligenceQwen3 30B A3B 2507 Thinking by 2.67.29.8+2.6
- IFEvalLFM2.5-8B-A1B by 191.890.8+1
- BFCLv4tie49.750.5tie
Questions people ask
Which is better, LFM2.5-8B-A1B or Qwen3 30B A3B 2507 Thinking?
Qwen3 30B A3B 2507 Thinking wins four of the seven areas where both have results: coding, reasoning, math and long documents. LFM2.5-8B-A1B wins following instructions. They are level on agents and facts.
Which is better for coding?
Qwen3 30B A3B 2507 Thinking. It wins 1 of the 2 coding tests both models report; LFM2.5-8B-A1B wins none, and 1 is a tie.
How do you compare the two?
We use the 21 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 10 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.