Skip to main content
BenchLM
Data
Recommendation
Ranking data as of

Best Value LLM for Coding in 2026 — Cost-Adjusted Rankings

As of October 29, 2026, the top model among value LLM for coding on the BenchLM leaderboard is Ministral 3 3B with a score of 281.6.

Bottom line: value here is coding score per output dollar, so check each row's raw score before you shortlist it.

1Open weights

Ministral 3 3B

Mistral · 128K

281.6Score/$

Score: 28.2 · $0.1/1M

2Open weights

MiMo-V2.6-Flash

Xiaomi · 1M

198.82Score/$

Score: 55.7 · $0.28/1M

3Open weights

Ministral 3 8B

Mistral · 128K

193.67Score/$

Score: 29.1 · $0.15/1M

Full Rankings (96 ranked, 1 tracked)

1
Ministral 3 3B
Mistral·Open Weight·128K

281.6

Score/$

Score: 28.2 · $0.1/1M

2
MiMo-V2.6-Flash
Xiaomi·Open Weight·1M

198.82

Score/$

Score: 55.7 · $0.28/1M

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

193.67

Score/$

Score: 29.1 · $0.15/1M

4
Ministral 3 14B
Mistral·Open Weight·128K

149.6

Score/$

Score: 29.9 · $0.2/1M

5
Mercury 2.5
Inception·Proprietary·260K

141.67

Score/$

Score: 21.3 · $0.15/1M

6
Claude Haiku 5.5
Anthropic·Proprietary·1M

104.84

Score/$

Score: 52.4 · $0.5/1M

7
GPT-6 Luna
OpenAI·Proprietary·1.05M

102.8

Score/$

Score: 51.4 · $0.5/1M

8
MiniMax M3.1 Flash
MiniMax·Proprietary·1M

89.17

Score/$

Score: 53.5 · $0.6/1M

9
DeepSeek V3.2
DeepSeek·Open Weight·128K

69.55

Score/$

Score: 29.2 · $0.42/1M

10
Qwen3.5 Flash
Alibaba·Proprietary·1M

67.82

Score/$

Score: 27.1 · $0.4/1M

11
MiMo-V2.6-Pro
Xiaomi·Open Weight·1M

65.4

Score/$

Score: 56.9 · $0.87/1M

12
DeepSeek V4.1 Flash 1017
DeepSeek·Open Weight·1M

52.43

Score/$

Score: 62.9 · $1.2/1M

13
GPT-5.6 Luna
OpenAI·Proprietary·1.05M

51.45

Score/$

Score: 61.7 · $1.2/1M

14
Mistral Small 4
Mistral·Open Weight·256K

43.17

Score/$

Score: 25.9 · $0.6/1M

15
MiniMax M3
MiniMax·Open Weight·1M

31.14

Score/$

Score: 37.4 · $1.2/1M

16
MiniMax M2.7
MiniMax·Open Weight·200K

28.94

Score/$

Score: 34.7 · $1.2/1M

17
MiniMax M2.5
MiniMax·Proprietary·128K

28.73

Score/$

Score: 34.5 · $1.2/1M

18
GPT-4.1 nano
OpenAI·Proprietary·1M

28.62

Score/$

Score: 11.5 · $0.4/1M

19
Step 3.7 Flash
StepFun·Open Weight·256K

28.08

Score/$

Score: 32.3 · $1.15/1M

20
Celeris-1
Celeris·Proprietary·128K

27.51

Score/$

Score: 19.3 · $0.7/1M

21
Inkling-Small
Thinking Machines Lab·Open Weight·1M

25.62

Score/$

Score: 36.9 · $1.44/1M

22
GPT-5.4 nano
OpenAI·Proprietary·400K

24.09

Score/$

Score: 30.1 · $1.25/1M

23
Trinity-Large-Thinking
Arcee AI·Open Weight·512K

22.01

Score/$

Score: 19.8 · $0.9/1M

24
Quasar 438B
Multiverse Computing·Proprietary·1M

21.74

Score/$

Score: 39.1 · $1.8/1M

25
Mercury 2
Inception·Proprietary·128K

21.69

Score/$

Score: 16.3 · $0.75/1M

26
Mistral Large 4
Mistral·Open Weight·1M

20.73

Score/$

Score: 43.5 · $2.1/1M

27
Gemini 3.1 Flash-Lite
Google·Proprietary·1M

18.85

Score/$

Score: 28.3 · $1.5/1M

28
Step 5 Preview
StepFun·Open Weight·1M

17.89

Score/$

Score: 48.3 · $2.7/1M

29
Gemini 3.8 Flash
Google·Proprietary·1M

15.95

Score/$

Score: 59.8 · $3.75/1M

30
Gemini 3.7 Flash
Google·Proprietary·1M

15.34

Score/$

Score: 57.5 · $3.75/1M

31
DeepSeek V3.2 (Thinking)
DeepSeek·Open Weight·128K

14.79

Score/$

Score: 32.4 · $2.19/1M

32
GPT-4o mini
OpenAI·Proprietary·128K

14.78

Score/$

Score: 8.9 · $0.6/1M

33
DeepSeek V3
DeepSeek·Open Weight·128K

14.37

Score/$

Score: 15.8 · $1.1/1M

34
Gemini 3.5 Flash-Lite
Google·Proprietary·1M

14.04

Score/$

Score: 35.1 · $2.5/1M

35
GPT-5 mini
OpenAI·Proprietary·128K

13.85

Score/$

Score: 27.7 · $2/1M

36
Gemini 3.6 Flash
Google·Proprietary·1M

13.64

Score/$

Score: 51.2 · $3.75/1M

37
Grok Build 0.1
xAI·Proprietary·256K

13.24

Score/$

Score: 26.5 · $2/1M

38
Kimi K2.5
Moonshot AI·Open Weight·256K

13.09

Score/$

Score: 39.3 · $3/1M

39
Muse Spark 1.2
Meta·Proprietary·1M

12.43

Score/$

Score: 52.8 · $4.25/1M

40
GLM-5.2
Z.AI·Open Weight·1M

12.2

Score/$

Score: 53.7 · $4.4/1M

41
GLM-5
Z.AI·Open Weight·200K

12.07

Score/$

Score: 38.6 · $3.2/1M

42
DeepSeek V4 Pro 0813
DeepSeek·Open Weight·1M

12.05

Score/$

Score: 47.7 · $3.96/1M

43
Grok Code Fast 1
xAI·Proprietary·256K

11.81

Score/$

Score: 17.7 · $1.5/1M

44
GPT-4.1 mini
OpenAI·Proprietary·1M

11.06

Score/$

Score: 17.7 · $1.6/1M

45
Kimi K2.6
Moonshot AI·Open Weight·256K

10.73

Score/$

Score: 42.9 · $4/1M

46
GLM-5.1
Z.AI·Open Weight·203K

10.7

Score/$

Score: 47.1 · $4.4/1M

47
Gemini 3 Flash
Google·Proprietary·1M

10.7

Score/$

Score: 32.1 · $3/1M

48
Grok 4.3
xAI·Proprietary·1M

10.66

Score/$

Score: 26.7 · $2.5/1M

49
Kimi K2.7 Code
Moonshot AI·Open Weight·256K

10.51

Score/$

Score: 42 · $4/1M

50
Mistral Large 3
Mistral·Proprietary·128K

10.39

Score/$

Score: 15.6 · $1.5/1M

51
Grok 4.6
xAI·Proprietary·500K

9.74

Score/$

Score: 58.4 · $6/1M

52
Grok 4.5
xAI·Proprietary·500K

9.25

Score/$

Score: 55.5 · $6/1M

53
Grok 4.20
xAI·Proprietary·2M

8.78

Score/$

Score: 22 · $2.5/1M

54
GLM-5V-Turbo
Z.AI·Proprietary·200K

8.2

Score/$

Score: 32.8 · $4/1M

55
Claude Sonnet 5.5
Anthropic·Proprietary·1M

8.17

Score/$

Score: 81.7 · $10/1M

56
Gemini 4 Carbon
Google·Proprietary·

8.11

Score/$

Score: 81.1 · $10/1M

57
GPT-5.4 mini
OpenAI·Proprietary·400K

7.85

Score/$

Score: 35.3 · $4.5/1M

58
Gemini 4 Argon
Google·Proprietary·1M

7.22

Score/$

Score: 72.2 · $10/1M

59
Inkling
Thinking Machines Lab·Open Weight·1M

7.04

Score/$

Score: 33 · $4.68/1M

60
GPT-6.1 Sol
OpenAI·Proprietary·1.05M

7.03

Score/$

Score: 70.3 · $10/1M

61
GPT-6 Sol
OpenAI·Proprietary·1.05M

6.5

Score/$

Score: 65 · $10/1M

62
Claude Sonnet 5
Anthropic·Proprietary·1M

5.75

Score/$

Score: 57.5 · $10/1M

63
Gemini 3.5 Flash
Google·Proprietary·1M

5.59

Score/$

Score: 50.4 · $9/1M

64
GPT-5.6 Terra
OpenAI·Proprietary·1.05M

5.17

Score/$

Score: 62 · $12/1M

65
Claude Opus 5.5
Anthropic·Proprietary·1M

4.14

Score/$

Score: 82.9 · $20/1M

66
Kimi K3
Moonshot AI·Open Weight·1.05M

3.91

Score/$

Score: 58.7 · $15/1M

67
GPT-5.3 Codex
OpenAI·Proprietary·400K

3.82

Score/$

Score: 53.5 · $14/1M

68
Gemini 3 Pro
Google·Proprietary·2M

3.64

Score/$

Score: 43.6 · $12/1M

69
GPT-5.1
OpenAI·Proprietary·200K

3.62

Score/$

Score: 36.2 · $10/1M

70
GPT-5.1-Codex
OpenAI·Proprietary·400K

3.58

Score/$

Score: 35.8 · $10/1M

71
Claude Haiku 4.5
Anthropic·Proprietary·200K

3.4

Score/$

Score: 17 · $5/1M

72
Mistral Medium 3.5 128B
Mistral·Open Weight·256K

3.39

Score/$

Score: 25.5 · $7.5/1M

73
GPT-5.6 Sol
OpenAI·Proprietary·1.05M

3.35

Score/$

Score: 67.1 · $20/1M

74
Muse Wonder
Meta·Proprietary·1M

3.29

Score/$

Score: 65.7 · $20/1M

75
Gemini 3.1 Pro
Google·Proprietary·1M

3.26

Score/$

Score: 39.1 · $12/1M

76
GPT-5.4
OpenAI·Proprietary·1.05M

3.19

Score/$

Score: 47.8 · $15/1M

77
GPT-5.2-Codex
OpenAI·Proprietary·400K

3.18

Score/$

Score: 44.5 · $14/1M

78
Gemini 1.5 Pro
Google·Proprietary·2M

3.02

Score/$

Score: 15.1 · $5/1M

79
Claude Sonnet 4.6
Anthropic·Proprietary·200K

2.98

Score/$

Score: 44.7 · $15/1M

80
Claude Opus 5
Anthropic·Proprietary·

2.82

Score/$

Score: 70.5 · $25/1M

81
GPT-5.2
OpenAI·Proprietary·400K

2.65

Score/$

Score: 37 · $14/1M

82
Claude Opus 4.8
Anthropic·Proprietary·1M

2.43

Score/$

Score: 60.7 · $25/1M

83
Gemini 2.5 Pro
Google·Proprietary·1M

2.27

Score/$

Score: 22.7 · $10/1M

84
Claude Opus 4.7 (Adaptive)
Anthropic·Proprietary·1M

2.21

Score/$

Score: 55.3 · $25/1M

85
Claude Opus 4.7
Anthropic·Proprietary·1M

2.2

Score/$

Score: 55.1 · $25/1M

86
GPT-5.5
OpenAI·Proprietary·1M

2.08

Score/$

Score: 62.3 · $30/1M

87
Claude Opus 4.6
Anthropic·Proprietary·1M

1.85

Score/$

Score: 46.1 · $25/1M

88
Claude Opus 4.5
Anthropic·Proprietary·200K

1.66

Score/$

Score: 41.5 · $25/1M

89
Claude Fable 5.1
Anthropic·Proprietary·1M

1.53

Score/$

Score: 76.4 · $50/1M

90
GPT-6 Astra
OpenAI·Proprietary·1.05M

1.49

Score/$

Score: 74.7 · $50/1M

91
Command A+
Cohere·Open Weight·128K

1.44

Score/$

Score: 14.4 · $10/1M

92
Claude Fable 5
Anthropic·Proprietary·1M+

1.43

Score/$

Score: 71.4 · $50/1M

93
o1
OpenAI·Proprietary·200K

0.48

Score/$

Score: 29.1 · $60/1M

94
GPT-4 Turbo
OpenAI·Proprietary·128K

0.36

Score/$

Score: 10.9 · $30/1M

95
o1-preview
OpenAI·Proprietary·200K

0.35

Score/$

Score: 21.1 · $60/1M

96
Claude 3 Opus
Anthropic·Proprietary·200K

0.16

Score/$

Score: 11.7 · $75/1M

—
Mistral Large 4
Mistral·Open Weight·1M

Open weights (mistralai/Mistral-Large-4-1T-A52B) under the Mistral Ecosystem Open-Weight license, a modified MIT license; businesses running a Model-as-a-Service offering need a separate commercial license. API ID mistral-large-4-0, with Low, Medium and High (default) reasoning modes.

—

tracked

How to choose

Maximum coding per dollar?Take the #1 value row above, then compare its raw coding score on the coding leaderboardStrong coding at good value?Take the cheapest row above whose raw coding score clears your barBest raw coding regardless of cost?See the standard coding leaderboard

Key Takeaways

The best value model is Ministral 3 3B by Mistral with a provisional Score/$ ratio of 281.6 (score: 28.2, output: $0.1/1M tokens).

The best open-weight model is Ministral 3 3B at position #1.

96 models are ranked here; 1 tracked model is listed without a position.

Score in Context

What these scores mean

Value scores divide the weighted coding score by output token price (per 1M tokens). Higher means more capability per dollar. Models with no listed price are excluded.

Known limitations

Value rankings favor cheap models even if absolute performance is modest. A model scoring half as well at one-tenth the price wins on value — but may not meet your quality bar. Always check raw scores alongside value rankings.

About this ranking

Ranking data as of October 29, 2026

Raw benchmark scores only tell half the story. This ranking divides each model's weighted coding score by its output token price, surfacing models that deliver the most coding capability per dollar spent. A model scoring 70 at $1/1M tokens outranks one scoring 80 at $15/1M tokens here — because for the same budget you get far more coding work done. Use this alongside the standard coding leaderboard to find the sweet spot between performance and cost for your coding workflows.

For a browser interface or a small web app, use the frontend model shortlist and test the generated UI before accepting it.

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

Ministral 3 3B leads this ranking with a score of 281.6, followed by MiMo-V2.6-Flash (198.82) and Ministral 3 8B (193.67). There is a significant gap between the leading models and the rest of the field.

The best open-weight option is Ministral 3 3B (ranked #1 with a score of 281.6). Open-weight models are highly competitive in this category — self-hosting is a viable alternative to proprietary APIs.

This ranking uses provisional weighted averages across the scoring benchmarks in coding. For detailed model profiles, click any model name above. To compare two specific models head-to-head, use the "vs #" links.

Questions

What is the best LLM for coding right now?

For raw coding capability, use the live coding leaderboard — it recomputes the order from calibrated coding evidence on every build, with evidence labels showing which positions are Supported versus Estimated. This page answers the follow-up question: which model delivers the most coding capability per dollar spent.

Is Claude or GPT better for coding?

The live coding leaderboard reflects the latest results. Evidence labels matter as much as rank: a Supported position slightly below an Estimated one is often the safer pick for production coding work.

What is the best open-source LLM for coding?

Use the open-weight rows on the coding leaderboard for capability order, and the open-source rankings for the full downloadable-weights picture. Open coding leaders trail the proprietary top tier on raw score but often win decisively on this page’s capability-per-dollar measure.

Which coding model gives the best value?

That is what this ranking measures: weighted coding score divided by output token price. A cheap model with a low coding score can rank high here, so read each row's raw coding score next to its price before you shortlist it.

Explore More

Last updated: October 29, 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 6,100+ subscribers.

One email each week. Unsubscribe anytime.