Follow model changes

Recommendation

Best Value Agentic AI Model in 2026 — Cost-Adjusted Rankings

As of October 1, 2026, the top model in best value agentic ai model on the BenchLM leaderboard is MiMo-V2.6-Flash with a score of 195.3.

Bottom line: Agentic tasks are token-intensive — value matters more here than almost anywhere.

Ranking data as of October 1, 2026

MiMo-V2.6-Flash

Xiaomi · 1M

Score: 54.7 · $0.28/1M

GPT-6 Luna

OpenAI · 1.05M

Score: 53 · $0.5/1M

MiMo-V2.6-Pro

Xiaomi · 1M

Score: 66.4 · $0.87/1M

Full Rankings (76 models)

1

MiMo-V2.6-Flash

Xiaomi·Open Weight·1M

195.29

Score/$

Score: 54.7 · $0.28/1M

2

OpenAI·Proprietary·1.05M

106.06

Score/$

Score: 53 · $0.5/1M

3

Xiaomi·Open Weight·1M

76.37

Score/$

Score: 66.4 · $0.87/1M

4

Inception·Proprietary·260K

75.13

Score/$

Score: 11.3 · $0.15/1M

5

DeepSeek V4.1 Flash

DeepSeek·Open Weight·1M

50.89

Score/$

Score: 61.1 · $1.2/1M

6

Mistral·Open Weight·128K

47.9

Score/$

Score: 4.8 · $0.1/1M

7

OpenAI·Proprietary·1.05M

45.88

Score/$

Score: 55.1 · $1.2/1M

8

Mistral·Open Weight·128K

36.53

Score/$

Score: 5.5 · $0.15/1M

9

MiniMax·Open Weight·1M

33.23

Score/$

Score: 39.9 · $1.2/1M

10

Ministral 3 14B

Mistral·Open Weight·128K

31.2

Score/$

Score: 6.2 · $0.2/1M

11

StepFun·Open Weight·256K

31.14

Score/$

Score: 35.8 · $1.15/1M

12

OpenAI·Proprietary·400K

25.45

Score/$

Score: 31.8 · $1.25/1M

13

Thinking Machines Lab·Open Weight·1M

24.53

Score/$

Score: 35.3 · $1.44/1M

14

MiniMax·Open Weight·200K

24.29

Score/$

Score: 29.2 · $1.2/1M

15

MiniMax·Proprietary·128K

23.81

Score/$

Score: 28.6 · $1.2/1M

16

StepFun·Pending·1M

23.51

Score/$

Score: 63.5 · $2.7/1M

17

Mistral Small 4

Mistral·Open Weight·256K

22.12

Score/$

Score: 13.3 · $0.6/1M

18

Multiverse Computing·Proprietary·1M

21.62

Score/$

Score: 38.9 · $1.8/1M

19

Inception·Proprietary·128K

19.8

Score/$

Score: 14.9 · $0.75/1M

20

Gemini 3.1 Flash-Lite

Google·Proprietary·1M

18.48

Score/$

Score: 27.7 · $1.5/1M

21

Gemini 3.8 Flash

Google·Proprietary·1M

17.41

Score/$

Score: 65.3 · $3.75/1M

22

Gemini 3.7 Flash

Google·Proprietary·1M

15.47

Score/$

Score: 58 · $3.75/1M

23

Trinity-Large-Thinking

Arcee AI·Open Weight·512K

14.09

Score/$

Score: 12.7 · $0.9/1M

24

Meta·Proprietary·1M

13.96

Score/$

Score: 59.4 · $4.25/1M

25

xAI·Proprietary·256K

13.93

Score/$

Score: 27.9 · $2/1M

26

OpenAI·Proprietary·128K

13.85

Score/$

Score: 27.7 · $2/1M

27

Gemini 3.5 Flash-Lite

Google·Proprietary·1M

13.65

Score/$

Score: 34.1 · $2.5/1M

28

DeepSeek V4 Pro 0813

DeepSeek·Open Weight·1M

13.38

Score/$

Score: 53 · $3.96/1M

29

Z.AI·Open Weight·1M

12.75

Score/$

Score: 56.1 · $4.4/1M

30

Moonshot AI·Open Weight·256K

12.65

Score/$

Score: 38 · $3/1M

31

Z.AI·Open Weight·200K

12.13

Score/$

Score: 38.8 · $3.2/1M

32

Gemini 3.6 Flash

Google·Proprietary·1M

11.99

Score/$

Score: 45 · $3.75/1M

33

xAI·Proprietary·500K

11.35

Score/$

Score: 68.1 · $6/1M

34

Moonshot AI·Open Weight·256K

10.84

Score/$

Score: 43.4 · $4/1M

35

Google·Proprietary·1M

10.57

Score/$

Score: 31.7 · $3/1M

36

xAI·Proprietary·1M

10.4

Score/$

Score: 26 · $2.5/1M

37

Z.AI·Open Weight·203K

9.61

Score/$

Score: 42.3 · $4.4/1M

38

xAI·Proprietary·2M

9.6

Score/$

Score: 24 · $2.5/1M

39

xAI·Proprietary·500K

9.42

Score/$

Score: 56.5 · $6/1M

40

Moonshot AI·Open Weight·256K

9.4

Score/$

Score: 37.6 · $4/1M

41

DeepSeek·Open Weight·128K

9.38

Score/$

Score: 10.3 · $1.1/1M

42

Celeris·Proprietary·128K

9.24

Score/$

Score: 6.5 · $0.7/1M

43

Thinking Machines Lab·Open Weight·1M

7.73

Score/$

Score: 36.2 · $4.68/1M

44

OpenAI·Proprietary·400K

7.57

Score/$

Score: 34.1 · $4.5/1M

45

Google·Proprietary·

7.42

Score/$

Score: 74.2 · $10/1M

46

Mistral Large 3

Mistral·Proprietary·128K

6.95

Score/$

Score: 10.4 · $1.5/1M

47

Claude Sonnet 5.5

Anthropic·Proprietary·1M

6.8

Score/$

Score: 68 · $10/1M

48

OpenAI·Proprietary·1.05M

6.78

Score/$

Score: 67.8 · $10/1M

49

Claude Sonnet 5

Anthropic·Proprietary·1M

6.46

Score/$

Score: 64.6 · $10/1M

50

OpenAI·Proprietary·1.05M

5.91

Score/$

Score: 59.1 · $10/1M

51

Gemini 3.5 Flash

Google·Proprietary·1M

5.6

Score/$

Score: 50.4 · $9/1M

52

OpenAI·Proprietary·1.05M

4.94

Score/$

Score: 59.2 · $12/1M

53

Moonshot AI·Pending·1.05M

4.52

Score/$

Score: 67.8 · $15/1M

54

Claude Opus 5.5

Anthropic·Proprietary·1M

4.4

Score/$

Score: 88.1 · $20/1M

55

Claude Haiku 4.5

Anthropic·Proprietary·200K

4.36

Score/$

Score: 21.8 · $5/1M

56

OpenAI·Proprietary·400K

3.96

Score/$

Score: 55.4 · $14/1M

57

OpenAI·Proprietary·1.05M

3.4

Score/$

Score: 68 · $20/1M

58

OpenAI·Proprietary·1.05M

3.3

Score/$

Score: 49.6 · $15/1M

59

Google·Proprietary·1M

3.2

Score/$

Score: 38.4 · $12/1M

60

Anthropic·Proprietary·

3.1

Score/$

Score: 77.6 · $25/1M

61

OpenAI·Proprietary·400K

3.06

Score/$

Score: 42.8 · $14/1M

62

Claude Sonnet 4.6

Anthropic·Proprietary·200K

2.9

Score/$

Score: 43.4 · $15/1M

63

OpenAI·Proprietary·400K

2.88

Score/$

Score: 40.3 · $14/1M

64

Mistral Medium 3.5 128B

Mistral·Open Weight·256K

2.57

Score/$

Score: 19.3 · $7.5/1M

65

Claude Opus 4.8

Anthropic·Proprietary·1M

2.46

Score/$

Score: 61.6 · $25/1M

66

Google·Proprietary·1M

2.44

Score/$

Score: 24.4 · $10/1M

67

Claude Opus 4.7 (Adaptive)

Anthropic·Proprietary·1M

2.37

Score/$

Score: 59.2 · $25/1M

68

Claude Sonnet 4.5

Anthropic·Proprietary·200K

2.26

Score/$

Score: 33.9 · $15/1M

69

Claude Opus 4.7

Anthropic·Proprietary·1M

2.13

Score/$

Score: 53.2 · $25/1M

70

OpenAI·Proprietary·1M

1.98

Score/$

Score: 59.5 · $30/1M

71

Claude Opus 4.6

Anthropic·Proprietary·1M

1.78

Score/$

Score: 44.4 · $25/1M

72

Cohere·Open Weight·128K

1.59

Score/$

Score: 15.9 · $10/1M

73

Claude Fable 5.1

Anthropic·Proprietary·1M

1.58

Score/$

Score: 78.9 · $50/1M

74

Anthropic·Proprietary·1M+

1.48

Score/$

Score: 74 · $50/1M

75

OpenAI·Proprietary·1.05M

1.41

Score/$

Score: 70.7 · $50/1M

76

Claude Opus 4.5

Anthropic·Proprietary·200K

1.24

Score/$

Score: 31.1 · $25/1M

How to choose

Cheapest agentic model?Take the cheapest row above whose raw agentic score clears your barAgentic value in OpenAI ecosystem?Compare current OpenAI token prices, then check agentic scoresBest raw agentic performance?See the standard agentic leaderboard

Key Takeaways

The best value model is MiMo-V2.6-Flash by Xiaomi with a provisional Score/$ ratio of 195.29 (score: 54.7, output: $0.28/1M tokens).

The best open-weight model is MiMo-V2.6-Flash at position #1.

76 models are included in this ranking.

Score in Context

What these scores mean

Value scores divide the weighted agentic 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 1, 2026

Agentic workloads are token-intensive — agents loop, retry, and chain multiple calls. That makes cost-per-token a critical factor alongside raw capability. This ranking divides each model's weighted agentic score (Terminal-Bench 2.0, BrowseComp, OSWorld-Verified) by its output token price. The result shows which models give you the most agent capability per dollar. If you're building production AI agents with budget constraints, this is where you start.

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

MiMo-V2.6-Flash leads this ranking with a score of 195.29, followed by GPT-6 Luna (106.06) and MiMo-V2.6-Pro (76.37). There is a significant gap between the leading models and the rest of the field.

The best open-weight option is MiMo-V2.6-Flash (ranked #1 with a score of 195.29). 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 agentic. 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: October 1, 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.