Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start free brief
BenchLM

BenchLM recommendation

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

Data verified

As of August 15, 2026, the top model in best value llm for coding on the BenchLM leaderboard is Laguna S 2.1 with a score of 268.6.

Last verified: August 15, 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.

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

Bottom line: Gemini 3.1 Flash-Lite dominates coding value at $0.40/1M output. DeepSeek Coder 2.0 offers the best absolute coding performance per dollar among serious coding models.

Laguna S 2.1 leads this ranking with a score of 268.6, followed by Ministral 3 3B (209.8) and Ministral 3 14B (167.65). There is a significant gap between the leading models and the rest of the field.

The best open-weight option is Laguna S 2.1 (ranked #1 with a score of 268.6). 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 below. To compare two specific models head-to-head, use the "vs #" links.

1Open

Laguna S 2.1

Poolside · 1M

268.6Score/$

Score: 53.7 · $0.2/1M

2Open

Ministral 3 3B

Mistral · 128K

209.8Score/$

Score: 21 · $0.1/1M

3Open

Ministral 3 14B

Mistral · 128K

167.65Score/$

Score: 33.5 · $0.2/1M

What changed

Gemini 3.1 Flash-Lite leads coding value — most coding capability per dollar spent.

DeepSeek Coder 2.0 best value among dedicated coding models with strong raw scores.

Gemini 2.5 Flash good coding value with broader model capabilities.

How to choose

Maximum coding per dollar?

Gemini 3.1 Flash-Lite — lowest cost with decent coding

Strong coding at good value?

DeepSeek Coder 2.0 — best raw coding scores per dollar

Best raw coding regardless of cost?

See the standard coding leaderboard

Full Rankings (84 models)

1
Laguna S 2.1
Poolside·Open Weight·1M

268.6

Score/$

Score: 53.7 · $0.2/1M

2
Ministral 3 3B
Mistral·Open Weight·128K

209.8

Score/$

Score: 21 · $0.1/1M

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

167.65

Score/$

Score: 33.5 · $0.2/1M

4
Ministral 3 8B
Mistral·Open Weight·128K

157.2

Score/$

Score: 23.6 · $0.15/1M

5
DeepSeek V3.2
DeepSeek·Open Weight·128K

118.79

Score/$

Score: 49.9 · $0.42/1M

6
GPT-4.1 nano
OpenAI·Proprietary·1M

106.05

Score/$

Score: 42.4 · $0.4/1M

7
Qwen3.5 Flash
Alibaba·Proprietary·1M

105.37

Score/$

Score: 42.2 · $0.4/1M

8
Mistral Small 4
Mistral·Open Weight·256K

77.72

Score/$

Score: 46.6 · $0.6/1M

9
GPT-5.6 Luna
OpenAI·Proprietary·1.05M

60.85

Score/$

Score: 73 · $1.2/1M

10
GPT-4o mini
OpenAI·Proprietary·128K

58.57

Score/$

Score: 35.1 · $0.6/1M

11
DeepSeek V4 Pro 0813
DeepSeek·Proprietary·1M

56.93

Score/$

Score: 49.5 · $0.87/1M

12
Step 3.7 Flash
StepFun·Open Weight·256K

43.91

Score/$

Score: 50.5 · $1.15/1M

13
MiniMax M2.5
MiniMax·Proprietary·128K

43.67

Score/$

Score: 52.4 · $1.2/1M

14
Inkling-Small
Thinking Machines Lab·Open Weight·1M

43.23

Score/$

Score: 62.3 · $1.44/1M

15
MiniMax M3
MiniMax·Open Weight·1M

38.94

Score/$

Score: 46.7 · $1.2/1M

16
DeepSeek V3
DeepSeek·Open Weight·128K

38.32

Score/$

Score: 42.2 · $1.1/1M

17
Mercury 2
Inception·Proprietary·128K

38.04

Score/$

Score: 28.5 · $0.75/1M

18
MiniMax M2.7
MiniMax·Open Weight·200K

36.83

Score/$

Score: 44.2 · $1.2/1M

19
Trinity-Large-Thinking
Arcee AI·Open Weight·512K

34.63

Score/$

Score: 31.2 · $0.9/1M

20
GPT-5.4 nano
OpenAI·Proprietary·400K

32.34

Score/$

Score: 40.4 · $1.25/1M

21
Gemini 3.1 Flash-Lite
Google·Proprietary·1M

27.96

Score/$

Score: 41.9 · $1.5/1M

22
GPT-4.1 mini
OpenAI·Proprietary·1M

27.96

Score/$

Score: 44.7 · $1.6/1M

23
DeepSeek V3.2 (Thinking)
DeepSeek·Open Weight·128K

23.38

Score/$

Score: 51.2 · $2.19/1M

24
Grok Code Fast 1
xAI·Proprietary·256K

22.35

Score/$

Score: 33.5 · $1.5/1M

25
Composer 2
Cursor·Proprietary·200K

21.06

Score/$

Score: 52.7 · $2.5/1M

26
Grok Build 0.1
xAI·Proprietary·256K

21.06

Score/$

Score: 42.1 · $2/1M

27
Gemini 3.5 Flash-Lite
Google·Proprietary·1M

21

Score/$

Score: 52.5 · $2.5/1M

28
GPT-5 mini
OpenAI·Proprietary·128K

19.93

Score/$

Score: 39.9 · $2/1M

29
Mistral Large 3
Mistral·Proprietary·128K

19.3

Score/$

Score: 29 · $1.5/1M

30
GLM-5
Z.AI·Open Weight·200K

18.96

Score/$

Score: 60.7 · $3.2/1M

31
Kimi K2.5
Moonshot AI·Open Weight·256K

18.83

Score/$

Score: 56.5 · $3/1M

32
Kimi K2.5 (Reasoning)
Moonshot AI·Proprietary·128K

18.09

Score/$

Score: 54.3 · $3/1M

33
Gemini 3.7 Flash
Google·Proprietary·1M

17.71

Score/$

Score: 66.4 · $3.75/1M

34
Qwen3.5 397B
Alibaba·Open Weight·128K

14.86

Score/$

Score: 53.5 · $3.6/1M

35
Gemini 3 Flash
Google·Proprietary·1M

14.84

Score/$

Score: 44.5 · $3/1M

36
Muse Spark 1.2
Meta·Proprietary·1M

14.78

Score/$

Score: 62.8 · $4.25/1M

37
GLM-5.2
Z.AI·Open Weight·1M

14.51

Score/$

Score: 63.8 · $4.4/1M

38
Grok 4.3
xAI·Proprietary·1M

14.48

Score/$

Score: 36.2 · $2.5/1M

39
GLM-5V-Turbo
Z.AI·Proprietary·200K

14.33

Score/$

Score: 57.3 · $4/1M

40
GLM-5.1
Z.AI·Open Weight·203K

13.97

Score/$

Score: 61.5 · $4.4/1M

41
Kimi K2.7 Code
Moonshot AI·Open Weight·256K

13.31

Score/$

Score: 53.2 · $4/1M

42
Kimi K2.6
Moonshot AI·Open Weight·256K

12.57

Score/$

Score: 50.3 · $4/1M

43
o3-mini
OpenAI·Proprietary·200K

10.72

Score/$

Score: 47.2 · $4.4/1M

44
Grok 4.6
xAI·Proprietary·500K

10.6

Score/$

Score: 63.6 · $6/1M

45
GPT-5.4 mini
OpenAI·Proprietary·400K

10.04

Score/$

Score: 45.2 · $4.5/1M

46
Grok 4.5
xAI·Proprietary·500K

9.55

Score/$

Score: 57.3 · $6/1M

47
Claude Haiku 4.5
Anthropic·Proprietary·200K

9.35

Score/$

Score: 46.7 · $5/1M

48
Inkling
Thinking Machines Lab·Open Weight·1M

9.03

Score/$

Score: 42.3 · $4.68/1M

49
Gemini 3.6 Flash
Google·Proprietary·1M

8.66

Score/$

Score: 65 · $7.5/1M

50
Grok 4.20
xAI·Proprietary·2M

7.69

Score/$

Score: 46.1 · $6/1M

51
Gemini 1.5 Pro
Google·Proprietary·2M

7.22

Score/$

Score: 36.1 · $5/1M

52
Claude Sonnet 5
Anthropic·Proprietary·1M

6.85

Score/$

Score: 68.5 · $10/1M

53
Gemini 3.5 Flash
Google·Proprietary·1M

6.78

Score/$

Score: 61 · $9/1M

54
GPT-4.1
OpenAI·Proprietary·1M

6.1

Score/$

Score: 48.8 · $8/1M

55
Mistral Medium 3.5 128B
Mistral·Open Weight·256K

5.9

Score/$

Score: 44.3 · $7.5/1M

56
GPT-5.6 Terra
OpenAI·Proprietary·1.05M

5.45

Score/$

Score: 65.4 · $12/1M

57
GPT-5.1
OpenAI·Proprietary·200K

5.32

Score/$

Score: 53.2 · $10/1M

58
Kimi K3
Moonshot AI·Pending·1.05M

5.19

Score/$

Score: 77.9 · $15/1M

59
Gemini 3 Pro
Google·Proprietary·2M

5.16

Score/$

Score: 61.9 · $12/1M

60
Command A+
Cohere·Open Weight·128K

4.72

Score/$

Score: 47.2 · $10/1M

61
GPT-5.3 Codex
OpenAI·Proprietary·400K

4.54

Score/$

Score: 63.5 · $14/1M

62
GPT-5.2-Codex
OpenAI·Proprietary·400K

4.1

Score/$

Score: 57.4 · $14/1M

63
Gemini 3.1 Pro
Google·Proprietary·1M

4.04

Score/$

Score: 48.5 · $12/1M

64
GPT-5.4
OpenAI·Proprietary·1.05M

3.92

Score/$

Score: 58.7 · $15/1M

65
GPT-5.2
OpenAI·Proprietary·400K

3.89

Score/$

Score: 54.4 · $14/1M

66
Claude Sonnet 4.6
Anthropic·Proprietary·200K

3.83

Score/$

Score: 57.5 · $15/1M

67
Gemini 2.5 Pro
Google·Proprietary·1M

3.45

Score/$

Score: 34.5 · $10/1M

68
Claude Sonnet 4.5
Anthropic·Proprietary·200K

3.44

Score/$

Score: 51.6 · $15/1M

69
Claude Opus 5
Anthropic·Proprietary·

3.12

Score/$

Score: 78.1 · $25/1M

70
Claude 4 Sonnet
Anthropic·Proprietary·200K

2.88

Score/$

Score: 43.2 · $15/1M

71
Claude Opus 4.8
Anthropic·Proprietary·1M

2.85

Score/$

Score: 71.2 · $25/1M

72
Claude Opus 4.7
Anthropic·Proprietary·1M

2.7

Score/$

Score: 67.4 · $25/1M

73
GPT-5.6 Sol
OpenAI·Proprietary·1.05M

2.62

Score/$

Score: 78.7 · $30/1M

74
Claude Opus 4.7 (Adaptive)
Anthropic·Proprietary·1M

2.61

Score/$

Score: 65.3 · $25/1M

75
Claude Opus 4.5
Anthropic·Proprietary·200K

2.48

Score/$

Score: 61.9 · $25/1M

76
Claude Opus 4.6
Anthropic·Proprietary·1M

2.47

Score/$

Score: 61.7 · $25/1M

77
GPT-5.5
OpenAI·Proprietary·1M

2.37

Score/$

Score: 71 · $30/1M

78
Claude Mythos 5
Anthropic·Proprietary·1M+

1.62

Score/$

Score: 81.1 · $50/1M

79
Claude Fable 5
Anthropic·Proprietary·1M+

1.62

Score/$

Score: 80.8 · $50/1M

80
GPT-4 Turbo
OpenAI·Proprietary·128K

1.01

Score/$

Score: 30.2 · $30/1M

81
o1
OpenAI·Proprietary·200K

0.83

Score/$

Score: 49.8 · $60/1M

82
o1-preview
OpenAI·Proprietary·200K

0.78

Score/$

Score: 47.1 · $60/1M

83
Claude 4.1 Opus
Anthropic·Proprietary·200K

0.61

Score/$

Score: 46 · $75/1M

84
Claude 3 Opus
Anthropic·Proprietary·200K

0.52

Score/$

Score: 38.9 · $75/1M

Key Takeaways

The best value model is Laguna S 2.1 by Poolside with a provisional Score/$ ratio of 268.6 (score: 53.7, output: $0.2/1M tokens).

The best open-weight model is Laguna S 2.1 at position #1.

84 models are included in this ranking.

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.

Explore More

Last updated: August 15, 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.