Model comparison
LFM2.5-8B-A1B vs Qwen3.5 397B
Head-to-head evidence from 14 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Verified leaderboard positions: LFM2.5-8B-A1B unranked; Qwen3.5 397B #20
BenchAlign evidence: LFM2.5-8B-A1B estimated; Qwen3.5 397B estimated. Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. LFM2.5-8B-A1B and Qwen3.5 397B share 14 comparable benchmark results. 2 of 8 categories are comparable. 4 results are unique to LFM2.5-8B-A1B; 42 to Qwen3.5 397B.
Updated July 16, 2026- Shared results
- 14
- LFM2.5-8B-A1B only
- 4
- Qwen3.5 397B only
- 42
- Comparable categories
- 2 / 8
Pick Qwen3.5 397B if you want the stronger benchmark profile. LFM2.5-8B-A1B only becomes the better choice if you want the cheaper token bill or you want the stronger reasoning-first profile.
Confidence note. This is a partial-evidence comparison with 14 shared benchmark results across 6 evidence categories; 2 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
Qwen3.5 397B is clearly ahead on the provisional aggregate, 59 to 37. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
Qwen3.5 397B's sharpest advantage is in mathematics, where it averages 90.6 against 50. The single biggest benchmark swing on the page is AIME26, 50.0% to 93.3%.
Qwen3.5 397B is also the more expensive model on tokens at $0.60 input / $3.60 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for LFM2.5-8B-A1B. That is roughly Infinityx on output cost alone. LFM2.5-8B-A1B is the reasoning model in the pair, while Qwen3.5 397B is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | LFM2.5-8B-A1B | Δ | Qwen3.5 397B |
|---|---|---|---|
| Math | LFM2.5-8B-A1B50.0 | Margin→ 40.6 | Qwen3.5 397B90.6 |
| Inst. Following | LFM2.5-8B-A1B68.8 | Margin→ 23.8 | Qwen3.5 397B92.6 |
| Agentic | LFM2.5-8B-A1BNot measured | MarginNo overlap | Qwen3.5 397B56.5 |
| Coding | LFM2.5-8B-A1BNot measured | MarginNo overlap | Qwen3.5 397B66.5 |
| Reasoning | LFM2.5-8B-A1BNot measured | MarginNo overlap | Qwen3.5 397B63.2 |
| Knowledge | LFM2.5-8B-A1BNot measured | MarginNo overlap | Qwen3.5 397B56.9 |
| Multilingual | LFM2.5-8B-A1BNot measured | MarginNo overlap | Qwen3.5 397B84.7 |
| Multimodal | LFM2.5-8B-A1BNot measured | MarginNo overlap | Qwen3.5 397B79.6 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
AIME26
MathA 50.0%B 93.3%Winner: Qwen3.5 397BΔ 43.3AIME26: LFM2.5-8B-A1B scored 50.0%; Qwen3.5 397B scored 93.3%. Qwen3.5 397B wins this benchmark. - Source ↗
IFEval
Inst. FollowingA 91.8%B 92.6%Winner: Qwen3.5 397BΔ 0.8IFEval: LFM2.5-8B-A1B scored 91.8%; Qwen3.5 397B scored 92.6%. Qwen3.5 397B wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | LFM2.5-8B-A1B | Qwen3.5 397B | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | LFM2.5-8B-A1B$0 input / $0 output | Qwen3.5 397B$0.6 input / $3.6 output | LFM2.5-8B-A1B has the lower combined listed price. |
| Generation speedtokens per second | LFM2.5-8B-A1BNot available | Qwen3.5 397B96 tok/s | A complete speed comparison is not available. |
| First-answer latencyseconds to first token | LFM2.5-8B-A1BNot available | Qwen3.5 397B2.44 s | A complete latency comparison is not available. |
| Context windowmaximum listed tokens | LFM2.5-8B-A1B128K | Qwen3.5 397B128K | Listed context windows are equal. |
Benchmark Deep Dive
Agentic19 benchmarks
| Benchmark | LFM2.5-8B-A1B | Qwen3.5 397B | Result |
|---|---|---|---|
| BFCL v4Source | 49.7% | — | Not comparable |
| τ²-bench resultsSource | 16.1% | 95.6% | Qwen3.5 397B leads |
| Terminal-Bench 2.0Source | — | 52.5% | Not comparable |
| BrowseCompSource | — | 62% | Not comparable |
| Claw-EvalSource | — | 56.8% | Not comparable |
| QwenClawBenchSource | — | 51.8% | Not comparable |
| τ³-bench resultsSource | — | 68.4% | Not comparable |
| VITA-BenchSource | — | 43.7% | Not comparable |
| DeepPlanningSource | — | 37.6% | Not comparable |
| ToolathlonSource | — | 36.3% | Not comparable |
| MCP AtlasSource | — | 46.1% | Not comparable |
| MCP-TasksSource | — | 74.2% | Not comparable |
| WideResearchSource | — | 74.0% | Not comparable |
| Gert LabsSource | — | 46.76% | Not comparable |
| ResearchClawBenchSource | — | 14.2% | Not comparable |
| AA Agentic IndexSource | — | 19.9% | Not comparable |
| APEX-Agents-AASource | — | 15.3% | Not comparable |
| GDPval-AASource | — | 23.1% | Not comparable |
| GDPval-AASource | — | 962 | Not comparable |
Coding6 benchmarks
| Benchmark | LFM2.5-8B-A1B | Qwen3.5 397B | Result |
|---|---|---|---|
| Terminal-Bench HardSource | 4.5% | 40.9% | Qwen3.5 397B leads |
| AA-SciCodeSource | 7.8% | 42.0% | Qwen3.5 397B leads |
| SWE-bench VerifiedSource | — | 76.2% | Not comparable |
| LiveCodeBench v6Source | — | 83.6% | Not comparable |
| SWE-bench ProSource | — | 50.9% | Not comparable |
| AA Coding IndexSource | — | 48.2% | Not comparable |
Reasoning4 benchmarks
Knowledge12 benchmarks
| Benchmark | LFM2.5-8B-A1B | Qwen3.5 397B | Result |
|---|---|---|---|
| AA-GPQA DiamondSource | 51.3% | 89.3% | Qwen3.5 397B leads |
| AA-HLESource | 6.9% | 27.3% | Qwen3.5 397B leads |
| AA-Omniscience IndexSource | -33.3% | -29.8% | Qwen3.5 397B leads |
| AA-Omniscience AccuracySource | 9.4% | 31.4% | Qwen3.5 397B leads |
| AA-Omniscience Hallucination RateSource | 47.0% | 89.1% | LFM2.5-8B-A1B leads |
| Artificial Analysis Intelligence IndexSource | 8.3% | 33.7% | Qwen3.5 397B leads |
| GPQASource | — | 88.4% | Not comparable |
| SuperGPQASource | — | 70.4% | Not comparable |
| MMLU-ProSource | — | 87.8% | Not comparable |
| MMLU-ReduxSource | — | 94.9% | Not comparable |
| C-EvalSource | — | 93% | Not comparable |
| HLESource | — | 28.7% | Not comparable |
MathQwen3.5 397B wins7 benchmarks
| Benchmark | LFM2.5-8B-A1B | Qwen3.5 397B | Result |
|---|---|---|---|
| MATH-500Source | 88.8% | — | Not comparable |
| AIME 2025Source | 42.5% | — | Not comparable |
| AIME26Source | 50.0% | 93.3% | Qwen3.5 397B leads |
| HMMT Feb 2025Source | — | 94.8% | Not comparable |
| HMMT Nov 2025Source | — | 92.7% | Not comparable |
| HMMT Feb 2026Source | — | 87.9% | Not comparable |
| MMAnswerBenchSource | — | 80.9% | Not comparable |
Multilingual2 benchmarks
Multimodal7 benchmarks
Frequently Asked Questions (3)
Which is better, LFM2.5-8B-A1B or Qwen3.5 397B?
Qwen3.5 397B is ahead on BenchLM's provisional leaderboard, 59 to 37. The biggest single separator in this matchup is AIME26, where the scores are 50.0% and 93.3%.
Which is better for math, LFM2.5-8B-A1B or Qwen3.5 397B?
Qwen3.5 397B has the edge for math in this comparison, averaging 90.6 versus 50. Inside this category, AIME26 is the benchmark that creates the most daylight between them.
Which is better for instruction following, LFM2.5-8B-A1B or Qwen3.5 397B?
Qwen3.5 397B has the edge for instruction following in this comparison, averaging 92.6 versus 68.8. Inside this category, AA-IFBench is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
The AI models change fast. We track them for you.
A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.
Free. No spam. Unsubscribe anytime.