Skip to main content
BenchLM
Data
Recommendation
Ranking data as of

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

As of November 16, 2026, the top model among value agentic AI model on the BenchLM leaderboard is Ministral 3 3B with a score of 312.5.

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

Full Rankings (91 ranked)

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

312.5

Score/$

Score: 31.3 · $0.1/1M

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

221.32

Score/$

Score: 62 · $0.28/1M

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

208

Score/$

Score: 31.2 · $0.15/1M

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

156.6

Score/$

Score: 31.3 · $0.2/1M

vs #5
5
Claude Haiku 5.5
Anthropic·Proprietary·1M

118.42

Score/$

Score: 59.2 · $0.5/1M

vs #6
6
GPT-6 Luna
OpenAI·Proprietary·1.05M

109.32

Score/$

Score: 54.7 · $0.5/1M

vs #7
7
Muse Ember
Meta·Proprietary·1M

89.94

Score/$

Score: 45 · $0.5/1M

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

89.8

Score/$

Score: 53.9 · $0.6/1M

vs #9
9
Mercury 2.5
Inception·Proprietary·260K

87.8

Score/$

Score: 13.2 · $0.15/1M

vs #10
10
MiMo-V2.6-Pro
Xiaomi·Open Weight·1M

71.85

Score/$

Score: 62.5 · $0.87/1M

vs #11
11
Inkling 2 Small
Thinking Machines Lab·Proprietary·1M

55.69

Score/$

Score: 55.7 · $1/1M

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

51.26

Score/$

Score: 61.5 · $1.2/1M

vs #13
13
GPT-5.6 Luna
OpenAI·Proprietary·1.05M

47.78

Score/$

Score: 57.3 · $1.2/1M

vs #14
14
Step 3.7 Flash
StepFun·Open Weight·256K

34.42

Score/$

Score: 39.6 · $1.15/1M

vs #15
15
MiniMax M3
MiniMax·Open Weight·1M

34.33

Score/$

Score: 41.2 · $1.2/1M

vs #16
16
LFM2.5-24B-A2B
LiquidAI·Open Weight·131K

29.96

Score/$

Score: 30 · $1/1M

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

27.13

Score/$

Score: 32.6 · $1.2/1M

vs #18
18
Inkling-Small
Thinking Machines Lab·Open Weight·1M

26.13

Score/$

Score: 37.6 · $1.44/1M

vs #19
19
Celeris-1
Celeris·Proprietary·128K

25.97

Score/$

Score: 18.2 · $0.7/1M

vs #20
20
Trinity-Large-Thinking
Arcee AI·Open Weight·512K

25.94

Score/$

Score: 23.4 · $0.9/1M

vs #21
21
GPT-5.4 nano
OpenAI·Proprietary·400K

25.6

Score/$

Score: 32 · $1.25/1M

vs #22
22
MiniMax M2.5
MiniMax·Proprietary·128K

25.32

Score/$

Score: 30.4 · $1.2/1M

vs #23
23
Mistral Large 4
Mistral·Open Weight·1M

25.05

Score/$

Score: 52.6 · $2.1/1M

vs #24
24
Quasar 438B
Multiverse Computing·Proprietary·1M

24.62

Score/$

Score: 44.3 · $1.8/1M

vs #25
25
Mercury 2
Inception·Proprietary·128K

23

Score/$

Score: 17.3 · $0.75/1M

vs #26
26
Gemini 3.1 Flash-Lite
Google·Proprietary·1M

20.29

Score/$

Score: 30.4 · $1.5/1M

vs #27
27
Step 5 Preview
StepFun·Open Weight·1M

20.16

Score/$

Score: 54.4 · $2.7/1M

vs #28
28
Mistral Small 4
Mistral·Open Weight·256K

19.6

Score/$

Score: 11.8 · $0.6/1M

vs #29
29
Grok Build 0.1
xAI·Proprietary·256K

17.3

Score/$

Score: 34.6 · $2/1M

vs #30
30
Gemini 3.8 Flash
Google·Proprietary·1M

16.92

Score/$

Score: 63.4 · $3.75/1M

vs #31
31
Gemini 3.7 Flash
Google·Proprietary·1M

15.9

Score/$

Score: 59.6 · $3.75/1M

vs #32
32
GPT-5 mini
OpenAI·Proprietary·128K

15.47

Score/$

Score: 30.9 · $2/1M

vs #33
33
GPT-6.1 Terra
OpenAI·Proprietary·1.05M

15.12

Score/$

Score: 75.6 · $5/1M

vs #34
34
Qwen4 Max
Alibaba·Proprietary·1M

14.77

Score/$

Score: 66.5 · $4.5/1M

vs #35
35
DeepSeek V3
DeepSeek·Open Weight·128K

14.09

Score/$

Score: 15.5 · $1.1/1M

vs #36
36
Gemini 3.6 Flash
Google·Proprietary·1M

13.91

Score/$

Score: 52.2 · $3.75/1M

vs #37
37
DeepSeek V4 Pro 0813
DeepSeek·Open Weight·1M

13.7

Score/$

Score: 54.3 · $3.96/1M

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

13.7

Score/$

Score: 41.1 · $3/1M

vs #39
39
Gemini 3.5 Flash-Lite
Google·Proprietary·1M

13.66

Score/$

Score: 34.2 · $2.5/1M

vs #40
40
Muse Spark 1.2
Meta·Proprietary·1M

13.58

Score/$

Score: 57.7 · $4.25/1M

vs #41
41
GLM-5
Z.AI·Open Weight·200K

13.17

Score/$

Score: 42.1 · $3.2/1M

vs #42
42
Mistral Large 3
Mistral·Open Weight·128K

12.99

Score/$

Score: 19.5 · $1.5/1M

vs #43
43
GLM-5.2
Z.AI·Open Weight·1M

12.3

Score/$

Score: 54.1 · $4.4/1M

vs #44
44
Qwen3.5 397B
Alibaba·Open Weight·128K

12.14

Score/$

Score: 43.7 · $3.6/1M

vs #45
45
Grok 4.6
xAI·Proprietary·500K

11.33

Score/$

Score: 68 · $6/1M

vs #46
46
Kimi K2.5 (Reasoning)
Moonshot AI·Open Weight·128K

11.32

Score/$

Score: 34 · $3/1M

vs #47
47
Gemini 3 Flash
Google·Proprietary·1M

11.16

Score/$

Score: 33.5 · $3/1M

vs #48
48
Grok 4.3
xAI·Proprietary·1M

11.08

Score/$

Score: 27.7 · $2.5/1M

vs #49
49
Kimi K2.6
Moonshot AI·Open Weight·256K

10.93

Score/$

Score: 43.7 · $4/1M

vs #50
50
Gemini 4 Hydrogen
Google·Proprietary·1M

10.39

Score/$

Score: 77.9 · $7.5/1M

vs #51
51
Grok 4.20
xAI·Proprietary·2M

10.28

Score/$

Score: 25.7 · $2.5/1M

vs #52
52
GLM-5.1
Z.AI·Open Weight·203K

10.11

Score/$

Score: 44.5 · $4.4/1M

vs #53
53
Grok 4.5
xAI·Proprietary·500K

9.93

Score/$

Score: 59.6 · $6/1M

vs #54
54
Kimi K2.7 Code
Moonshot AI·Open Weight·256K

9.69

Score/$

Score: 38.8 · $4/1M

vs #55
55
Gemini 4 Carbon
Google·Proprietary·1M

8.65

Score/$

Score: 86.5 · $10/1M

vs #56
56
GPT-5.4 mini
OpenAI·Proprietary·400K

8.23

Score/$

Score: 37.1 · $4.5/1M

vs #57
57
Inkling
Thinking Machines Lab·Open Weight·1M

7.96

Score/$

Score: 37.3 · $4.68/1M

vs #58
58
Gemini 4 Argon
Google·Proprietary·1M

7.56

Score/$

Score: 75.6 · $10/1M

vs #59
59
Claude Sonnet 5.5
Anthropic·Proprietary·1M

7.31

Score/$

Score: 73.1 · $10/1M

vs #60
60
Kimi K3.1 Pro
Moonshot AI·Open Weight·1.05M

7.05

Score/$

Score: 70.5 · $10/1M

vs #61
61
GPT-6.1 Sol
OpenAI·Proprietary·1.05M

6.62

Score/$

Score: 66.2 · $10/1M

vs #62
62
GPT-6 Sol
OpenAI·Proprietary·1.05M

6.54

Score/$

Score: 65.4 · $10/1M

vs #63
63
Claude Sonnet 5
Anthropic·Proprietary·1M

6.25

Score/$

Score: 62.5 · $10/1M

vs #64
64
Gemini 3.5 Flash
Google·Proprietary·1M

5.58

Score/$

Score: 50.2 · $9/1M

vs #65
65
GPT-5.6 Terra
OpenAI·Proprietary·1.05M

4.9

Score/$

Score: 58.8 · $12/1M

vs #66
66
Claude Haiku 4.5
Anthropic·Proprietary·200K

4.49

Score/$

Score: 22.4 · $5/1M

vs #67
67
Kimi K3
Moonshot AI·Open Weight·1.05M

4.36

Score/$

Score: 65.5 · $15/1M

vs #68
68
Claude Opus 5.5
Anthropic·Proprietary·1M

4.21

Score/$

Score: 84.2 · $20/1M

vs #69
69
GPT-5.3 Codex
OpenAI·Proprietary·400K

3.81

Score/$

Score: 53.4 · $14/1M

vs #70
70
GPT-5.4
OpenAI·Proprietary·1.05M

3.44

Score/$

Score: 51.6 · $15/1M

vs #71
71
GPT-5.6 Sol
OpenAI·Proprietary·1.05M

3.33

Score/$

Score: 66.5 · $20/1M

vs #72
72
Muse Wonder
Meta·Proprietary·1M

3.32

Score/$

Score: 66.4 · $20/1M

vs #73
73
Gemini 3.1 Pro
Google·Proprietary·1M

3.3

Score/$

Score: 39.6 · $12/1M

vs #74
74
Claude Sonnet 4.6
Anthropic·Proprietary·200K

3.17

Score/$

Score: 47.5 · $15/1M

vs #75
75
GPT-5.2-Codex
OpenAI·Proprietary·400K

3.13

Score/$

Score: 43.8 · $14/1M

vs #76
76
GPT-5.2
OpenAI·Proprietary·400K

3.13

Score/$

Score: 43.9 · $14/1M

vs #77
77
Claude Opus 5
Anthropic·Proprietary·

3.02

Score/$

Score: 75.5 · $25/1M

vs #78
78
Gemini 2.5 Pro
Google·Proprietary·1M

2.94

Score/$

Score: 29.4 · $10/1M

vs #79
79
Mistral Medium 3.5 128B
Mistral·Open Weight·256K

2.65

Score/$

Score: 19.9 · $7.5/1M

vs #80
80
Claude Sonnet 4.5
Anthropic·Proprietary·200K

2.37

Score/$

Score: 35.6 · $15/1M

vs #81
81
Claude Opus 4.8
Anthropic·Proprietary·1M

2.31

Score/$

Score: 57.7 · $25/1M

vs #82
82
Claude Fable 5.2
Anthropic·Proprietary·1M

2.28

Score/$

Score: 91.2 · $40/1M

vs #83
83
Claude Opus 4.7 (Adaptive)
Anthropic·Proprietary·1M

2.22

Score/$

Score: 55.4 · $25/1M

vs #84
84
Claude Opus 4.7
Anthropic·Proprietary·1M

2.13

Score/$

Score: 53.1 · $25/1M

vs #85
85
GPT-5.5
OpenAI·Proprietary·1M

1.92

Score/$

Score: 57.7 · $30/1M

vs #86
86
Claude Opus 4.6
Anthropic·Proprietary·1M

1.85

Score/$

Score: 46.2 · $25/1M

vs #87
87
Claude Fable 5.1
Anthropic·Proprietary·1M

1.59

Score/$

Score: 79.3 · $50/1M

vs #88
88
GPT-6 Astra
OpenAI·Proprietary·1.05M

1.48

Score/$

Score: 73.8 · $50/1M

vs #89
89
Claude Fable 5
Anthropic·Proprietary·1M+

1.45

Score/$

Score: 72.7 · $50/1M

vs #90
90
Command A+
Cohere·Open Weight·128K

1.43

Score/$

Score: 14.3 · $10/1M

vs #91
91
Claude Opus 4.5
Anthropic·Proprietary·200K

1.1

Score/$

Score: 27.5 · $25/1M

How to choose

Key Takeaways

The best value model is Ministral 3 3B by Mistral with a provisional Score/$ ratio of 312.5 (score: 31.3, output: $0.1/1M tokens).

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

91 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 November 16, 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.

Ministral 3 3B leads this ranking with a score of 312.5, followed by MiMo-V2.6-Flash (221.32) and Ministral 3 8B (208). 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 312.5). 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.

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 7,000+ subscribers.

One email each week. Unsubscribe anytime.