Skip to main content
BenchLM
Recommendation

Best Non-Reasoning LLMs in 2026

As of September 22, 2026, the top model in best non-reasoning llms on the BenchLM leaderboard is Claude Opus 4.7 with a score of 66.4.

Bottom line: Non-reasoning models are faster and cheaper than chain-of-thought alternatives. Gemini 3.1 Pro leads this tier — proving that strong reasoning scores are possible without dedicated thinking tokens.

Data verified

Full Rankings (72 models)

1
Claude Opus 4.7
Anthropic·Proprietary·1M

66.4

BenchAlign v5.6

2
Claude Opus 4.6
Anthropic·Proprietary·1M

64.3

BenchAlign v5.6

3
Gemini 3 Pro
Google·Proprietary·2M

61.3

BenchAlign v5.6

4
Claude Sonnet 4.6
Anthropic·Proprietary·200K

56.7

BenchAlign v5.6

5
Inkling-Small
Thinking Machines Lab·Open Weight·1M

55.9

BenchAlign v5.6

6
Claude Opus 4.5
Anthropic·Proprietary·200K

55.9

BenchAlign v5.6

7
Gemini 3 Flash
Google·Proprietary·1M

55.6

BenchAlign v5.6

8
GLM-5
Z.AI·Open Weight·200K

55.4

BenchAlign v5.6

9
MiniMax M3
MiniMax·Open Weight·1M

55.3

BenchAlign v5.6

10
Inkling
Thinking Machines Lab·Open Weight·1M

54.4

BenchAlign v5.6

11
Kimi K2.5
Moonshot AI·Open Weight·256K

53.8

BenchAlign v5.6

12
Grok 4
xAI·Proprietary·128K

52.7

BenchAlign v5.6

13
Gemini 2.5 Pro
Google·Proprietary·1M

50.4

BenchAlign v5.6

14
DeepSeek V3.2
DeepSeek·Open Weight·128K

50.2

BenchAlign v5.6

15
MiniMax M2.5
MiniMax·Proprietary·128K

49.9

BenchAlign v5.6

16
GLM-5V-Turbo
Z.AI·Proprietary·200K

48.9

BenchAlign v5.6

17
MiniMax M2.7
MiniMax·Open Weight·200K

48.5

BenchAlign v5.6

18
Claude Sonnet 4.5
Anthropic·Proprietary·200K

47.9

BenchAlign v5.6

19
Gemini 3.1 Flash-Lite
Google·Proprietary·1M

47.5

BenchAlign v5.6

20
Grok 3 [Beta]
xAI·Proprietary·128K

45

BenchAlign v5.6

21
Gemini 2.5 Flash
Google·Proprietary·1M

43

BenchAlign v5.6

22
Claude Haiku 4.5
Anthropic·Proprietary·200K

41.6

BenchAlign v5.6

23
DeepSeek V3.1
DeepSeek·Open Weight·128K

41.2

BenchAlign v5.6

24
Step 3.5 Flash
StepFun·Open Weight·256K

41.1

BenchAlign v5.6

25
GPT-4.1
OpenAI·Proprietary·1M

40

BenchAlign v5.6

26
Claude 4.1 Opus
Anthropic·Proprietary·200K

38.5

BenchAlign v5.6

27
GPT-OSS 120B
OpenAI·Open Weight·128K

37.7

BenchAlign v5.6

28
Grok 4.1 Fast
xAI·Proprietary·1M

36.3

BenchAlign v5.6

29
Claude 4 Sonnet
Anthropic·Proprietary·200K

36.1

BenchAlign v5.6

30
GLM-4.5-Air
Z.AI·Proprietary·128K

34.6

BenchAlign v5.6

31
Mistral Small 4
Mistral·Open Weight·256K

34.3

BenchAlign v5.6

32
Mistral Large 3
Mistral·Proprietary·128K

34.1

BenchAlign v5.6

33
Ling 2.6 Flash
InclusionAI·Open Weight·262K

33.7

BenchAlign v5.6

34
GPT-OSS 20B
OpenAI·Open Weight·128K

33.6

BenchAlign v5.6

35
Grok Code Fast 1
xAI·Proprietary·256K

32

BenchAlign v5.6

36
DeepSeek V3
DeepSeek·Open Weight·128K

31.8

BenchAlign v5.6

37
Qwen2.5-72B
Alibaba·Open Weight·128K

31.3

BenchAlign v5.6

38
Qwen3-Omni-30B-A3B-Instruct
Alibaba·Open Weight·N/A

31.1

BenchAlign v5.6

39
Llama 3.1 405B
Meta·Open Weight·128K

30.8

BenchAlign v5.6

40
GPT-4o
OpenAI·Proprietary·128K

30.7

BenchAlign v5.6

41
Exaone 4.0 1.2B
LG AI Research·Open Weight·128K

30.5

BenchAlign v5.6

42
Granite-4.0-H-1B
IBM·Open Weight·128K

30.5

BenchAlign v5.6

43
Granite-4.0-350M
IBM·Open Weight·32K

30.2

BenchAlign v5.6

44
Granite-4.0-H-350M
IBM·Open Weight·32K

30.2

BenchAlign v5.6

45
Claude 3.5 Sonnet
Anthropic·Proprietary·200K

29.8

BenchAlign v5.6

46
Llama 4 Scout
Meta·Open Weight·10M

29.3

BenchAlign v5.6

47
Mistral Large 2
Mistral·Proprietary·128K

29

BenchAlign v5.6

48
Gemma 3 27B
Google·Open Weight·32K

28.7

BenchAlign v5.6

49
GPT-4.1 mini
OpenAI·Proprietary·1M

28.7

BenchAlign v5.6

50
Claude 3 Opus
Anthropic·Proprietary·200K

28.6

BenchAlign v5.6

51
Mistral Medium 3
Mistral·Proprietary·128K

28.3

BenchAlign v5.6

52
Gemini 1.5 Pro
Google·Proprietary·2M

27.9

BenchAlign v5.6

53
Qwen2.5 Coder 32B Instruct
Alibaba·Open Weight·128K

27.1

BenchAlign v5.6

54
GPT-4o mini
OpenAI·Proprietary·128K

26.8

BenchAlign v5.6

55
Kimi K2
Moonshot AI·Proprietary·128K

26.4

BenchAlign v5.6

56
Phi-4
Microsoft·Open Weight·16K

25.9

BenchAlign v5.6

57
Ministral 3 14B
Mistral·Open Weight·128K

25.2

BenchAlign v5.6

58
MiniMax M1 80k
MiniMax·Proprietary·80K

24.9

BenchAlign v5.6

59
GPT-4.1 nano
OpenAI·Proprietary·1M

24.3

BenchAlign v5.6

60
Llama 3 70B
Meta·Open Weight·128K

23.6

BenchAlign v5.6

61
Llama 4 Maverick
Meta·Open Weight·1M

22.9

BenchAlign v5.6

62
GPT-4 Turbo
OpenAI·Proprietary·128K

21.6

BenchAlign v5.6

63
Mixtral 8x22B Instruct v0.1
Mistral·Open Weight·64K

21.1

BenchAlign v5.6

64
Nova Pro
Amazon·Proprietary·128K

20.8

BenchAlign v5.6

65
LFM2-24B-A2B
LiquidAI·Proprietary·32K

19.9

BenchAlign v5.6

66
Ministral 3 8B
Mistral·Open Weight·128K

19.3

BenchAlign v5.6

67
LFM2.5-1.2B-Instruct
LiquidAI·Proprietary·32K

18.3

BenchAlign v5.6

68
Ministral 3 3B
Mistral·Open Weight·128K

18

BenchAlign v5.6

69
Claude 3 Haiku
Anthropic·Proprietary·200K

14.7

BenchAlign v5.6

70
Gemini 1.0 Pro
Google·Proprietary·32K

12.9

BenchAlign v5.6

71
Nemotron-4 15B
NVIDIA·Open Weight·32K

11.4

BenchAlign v5.6

72
Mistral 7B v0.3
Mistral·Open Weight·32K

8.1

BenchAlign v5.6

What changed

Gemini 3.1 Pro leads non-reasoning models — best reasoning (97), knowledge (96), and multilingual (100).

Claude Opus 4.6 most consistent non-reasoning model across all 8 categories.

Claude Sonnet 4.6 strong mid-tier with best multimodal (95) in this tier.

How to choose

Key Takeaways

The top model is Claude Opus 4.7 by Anthropic with a BenchAlign v5.6 score of 66.4 and Estimated evidence.

The best open-weight model is Inkling-Small at position #5.

72 models are included in this ranking.

Score in Context

What these scores mean

Non-reasoning models are standard completion/chat models without dedicated chain-of-thought. They are ranked by the same overall BenchLM score and are typically faster and cheaper per token.

Known limitations

The "non-reasoning" label excludes models with explicit chain-of-thought (like o3, DeepSeek R1). Some non-reasoning models still reason internally — the distinction is about architecture and pricing, not capability.

About this ranking

Last verified: September 22, 2026

Top standard AI models (no chain-of-thought reasoning) ranked by benchmark performance. Faster and cheaper than reasoning models.

Unless noted otherwise, ranking surfaces on this page use BenchLM’s provisional leaderboard lane rather than the stricter sourced-only verified leaderboard.

Claude Opus 4.7 leads this ranking with a score of 66.4, followed by Claude Opus 4.6 (64.3) and Gemini 3 Pro (61.3). There is meaningful separation between the top models, suggesting genuine performance differences.

The best open-weight option is Inkling-Small (ranked #5 with a score of 55.9). While proprietary models lead, open-weight options are within striking distance for teams willing to trade a few points of performance for full model control.

This ranking uses provisional overall weighted scores from the active scoring formula. For detailed model profiles, click any model name above. To compare two specific models head-to-head, use the "vs #" links.

Explore More

Last updated: September 22, 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.