Modex
ModelsBenchmarksCompareRankingsAgent ToolsBlog
Built on public AI model data.
MethodologyAgent Tools
Created byDiego Gómez González
← Back to Modex

Capability signals

Model Benchmarks

Independent and reproducible signals for comparing AI model capabilities.

Benchmarks are signals, not a universal truth. Keep source, methodology and coverage in view.

Benchmark dimension

Shows the mathematical trade-off boundary between lower input price and higher benchmark score. It is not an overall winner.

Intelligence against input price

The x-axis is logarithmic so free and paid models can share the same view. Each point represents one model with a source-reported score and input price.

Hover or focus a point
4.8014.9525.1035.2545.40$0.02$1.60$150.00QSOXATQOMSOMMACNNOOOIPMQONCQMLOUPZAPMNOOMAQNOMLMRMMMOOIGGML

Models in this signal — Intelligence

ModelScorePriceContext
QQwen: Qwen3.8 Max (0902)Qwen45.40$2.00out $6.001.0M
SStepFun: Step 5 PreviewStepfun43.70$1.00out $2.701.0M
OOpenAI: GPT-5.2-CodexOpenai28.50$1.75out $14.00400k
XASpaceXAI: Grok Build 0.1X Ai27.20$1.00out $2.00256k
TTencent: Hy3 previewTencent25.30$0.18out $0.60262k
QQwen: Qwen3.5-27BQwen22.90$0.27out $2.16262k
OOpenAI: o3 ProOpenai21.90$20.00out $80.00200k
MMeituan: LongCat 2.0Meituan19.10$0.30out $1.201.0M
SStepFun: Step 3.5 FlashStepfun17$0.10out $0.30262k
OOpenAI: o1Openai15.20$15.00out $60.00200k
MMistral: Mistral Medium 3.5Mistralai14.20$1.50out $7.50262k
MMistral: Mistral Medium 3.5 (batch)Mistralai14.20$0.75out $3.75262k
AAmazon: Nova 2 LiteAmazon13.40$0.30out $2.501.0M
CCohere: Command ACohere13.10$2.50out $10.00256k
NNVIDIA: Nemotron 3 SuperNvidia12.80$0.30out $2.40262k
NNVIDIA: Nemotron 3 Super (free)Nvidia12.80$0.00out $0.00262k
OOpenAI: o3 MiniOpenai12.50$1.10out $4.40200k
OOpenAI: o3 Mini (batch)Openai12.50$0.55out $2.20200k
OOpenAI: o1-proOpenai12.40$150.00out $600.00200k
IInception: Mercury 2.5Inception12.30$0.04out $0.15260k
PPerplexity: Sonar Reasoning ProPerplexity11.80$2.00out $8.00128k
MMiniMax: MiniMax M1Minimax11.70$0.55out $2.201.0M
QQwen: Qwen3 Next 80B A3B ThinkingQwen11.20$0.15out $1.20262k
OOpenAI: o3 Mini HighOpenai11$1.10out $4.40200k
NNVIDIA: Nemotron 3 Nano Omni (free)Nvidia10.30$0.00out $0.00256k
CCohere: North Mini Code (free)Cohere9.90$0.00out $0.00256k
QQwen: Qwen3 VL 30B A3B ThinkingQwen9.50$0.29out $1.00262k
MMistral: Devstral 2 2512Mistralai8.60$0.40out $2.00262k
LLiquidAI: LFM2.5-2.6B (free)Liquid8.40$0.00out $0.0066k
OOpenAI: GPT-4o (2024-11-20)Openai8.40$2.50out $10.00128k
UUpstage: Solar Pro 3Upstage7.80$0.15out $0.60131k
PPerplexity: SonarPerplexity7.70$1.00out $1.00127k
ZAZ.ai: GLM 4.5VZ Ai7.60$0.60out $1.8066k
PPerplexity: Sonar ProPerplexity7.60$3.00out $15.00200k
MMistral Large 2407Mistralai7.60$2.00out $6.00131k
NNous: Hermes 4 405BNousresearch7.50$1.00out $3.00131k
OOpenAI: GPT-4 TurboOpenai7$10.00out $30.00128k
OOpenAI: GPT-4 Turbo (batch)Openai7$5.00out $15.00128k
MMistral: Mistral Small 3Mistralai6.70$0.05out $0.0833k
AAmazon: Nova Lite 1.0Amazon6.70$0.06out $0.24300k
QQwen2.5 Coder 32B InstructQwen6.70$0.66out $1.0033k
NNous: Hermes 3 70B InstructNousresearch6.70$0.70out $0.70131k
OOpenAI: GPT-4Openai6.70$30.00out $60.008k
MLMeta: Llama 3.1 70B InstructMeta Llama6.60$0.40out $0.40131k
MMistral: SabaMistralai6.50$0.20out $0.6033k
RReka Flash 3Rekaai6.40$0.10out $0.2066k
MMicrosoft: Phi 4Microsoft5.90$0.07out $0.1416k
MMistral LargeMistralai5.80$2.00out $6.00128k
MMistral: Mixtral 8x22B InstructMistralai5.70$2.00out $6.0066k
OOpenAI: GPT-3.5 TurboOpenai5.50$0.50out $1.5016k
OOpenAI: GPT-3.5 Turbo (batch)Openai5.50$0.25out $0.7516k
IGIBM: Granite 4.0 MicroIbm Granite5.10$0.017out $0.112131k
GGoogle: Gemma 3 27BGoogle4.90$0.08out $0.35131k
MLMeta: Llama 3.2 1B InstructMeta Llama4.80$0.027out $0.20160k
Q

Qwen: Qwen3.8 Max (0902)

Qwen

45.40

Input

$2.00

Context

1.0M

S

StepFun: Step 5 Preview

Stepfun

43.70

Input

$1.00

Context

1.0M

O

OpenAI: GPT-5.2-Codex

Openai

28.50

Input

$1.75

Context

400k

XA

SpaceXAI: Grok Build 0.1

X Ai

27.20

Input

$1.00

Context

256k

T

Tencent: Hy3 preview

Tencent

25.30

Input

$0.18

Context

262k

Q

Qwen: Qwen3.5-27B

Qwen

22.90

Input

$0.27

Context

262k

O

OpenAI: o3 Pro

Openai

21.90

Input

$20.00

Context

200k

M

Meituan: LongCat 2.0

Meituan

19.10

Input

$0.30

Context

1.0M

S

StepFun: Step 3.5 Flash

Stepfun

17

Input

$0.10

Context

262k

O

OpenAI: o1

Openai

15.20

Input

$15.00

Context

200k

M

Mistral: Mistral Medium 3.5

Mistralai

14.20

Input

$1.50

Context

262k

M

Mistral: Mistral Medium 3.5 (batch)

Mistralai

14.20

Input

$0.75

Context

262k

A

Amazon: Nova 2 Lite

Amazon

13.40

Input

$0.30

Context

1.0M

C

Cohere: Command A

Cohere

13.10

Input

$2.50

Context

256k

N

NVIDIA: Nemotron 3 Super

Nvidia

12.80

Input

$0.30

Context

262k

N

NVIDIA: Nemotron 3 Super (free)

Nvidia

12.80

Input

$0.00

Context

262k

O

OpenAI: o3 Mini

Openai

12.50

Input

$1.10

Context

200k

O

OpenAI: o3 Mini (batch)

Openai

12.50

Input

$0.55

Context

200k

O

OpenAI: o1-pro

Openai

12.40

Input

$150.00

Context

200k

I

Inception: Mercury 2.5

Inception

12.30

Input

$0.04

Context

260k

P

Perplexity: Sonar Reasoning Pro

Perplexity

11.80

Input

$2.00

Context

128k

M

MiniMax: MiniMax M1

Minimax

11.70

Input

$0.55

Context

1.0M

Q

Qwen: Qwen3 Next 80B A3B Thinking

Qwen

11.20

Input

$0.15

Context

262k

O

OpenAI: o3 Mini High

Openai

11

Input

$1.10

Context

200k

N

NVIDIA: Nemotron 3 Nano Omni (free)

Nvidia

10.30

Input

$0.00

Context

256k

C

Cohere: North Mini Code (free)

Cohere

9.90

Input

$0.00

Context

256k

Q

Qwen: Qwen3 VL 30B A3B Thinking

Qwen

9.50

Input

$0.29

Context

262k

M

Mistral: Devstral 2 2512

Mistralai

8.60

Input

$0.40

Context

262k

L

LiquidAI: LFM2.5-2.6B (free)

Liquid

8.40

Input

$0.00

Context

66k

O

OpenAI: GPT-4o (2024-11-20)

Openai

8.40

Input

$2.50

Context

128k

U

Upstage: Solar Pro 3

Upstage

7.80

Input

$0.15

Context

131k

P

Perplexity: Sonar

Perplexity

7.70

Input

$1.00

Context

127k

ZA

Z.ai: GLM 4.5V

Z Ai

7.60

Input

$0.60

Context

66k

P

Perplexity: Sonar Pro

Perplexity

7.60

Input

$3.00

Context

200k

M

Mistral Large 2407

Mistralai

7.60

Input

$2.00

Context

131k

N

Nous: Hermes 4 405B

Nousresearch

7.50

Input

$1.00

Context

131k

O

OpenAI: GPT-4 Turbo

Openai

7

Input

$10.00

Context

128k

O

OpenAI: GPT-4 Turbo (batch)

Openai

7

Input

$5.00

Context

128k

M

Mistral: Mistral Small 3

Mistralai

6.70

Input

$0.05

Context

33k

A

Amazon: Nova Lite 1.0

Amazon

6.70

Input

$0.06

Context

300k

Q

Qwen2.5 Coder 32B Instruct

Qwen

6.70

Input

$0.66

Context

33k

N

Nous: Hermes 3 70B Instruct

Nousresearch

6.70

Input

$0.70

Context

131k

O

OpenAI: GPT-4

Openai

6.70

Input

$30.00

Context

8k

ML

Meta: Llama 3.1 70B Instruct

Meta Llama

6.60

Input

$0.40

Context

131k

M

Mistral: Saba

Mistralai

6.50

Input

$0.20

Context

33k

R

Reka Flash 3

Rekaai

6.40

Input

$0.10

Context

66k

M

Microsoft: Phi 4

Microsoft

5.90

Input

$0.07

Context

16k

M

Mistral Large

Mistralai

5.80

Input

$2.00

Context

128k

M

Mistral: Mixtral 8x22B Instruct

Mistralai

5.70

Input

$2.00

Context

66k

O

OpenAI: GPT-3.5 Turbo

Openai

5.50

Input

$0.50

Context

16k

O

OpenAI: GPT-3.5 Turbo (batch)

Openai

5.50

Input

$0.25

Context

16k

IG

IBM: Granite 4.0 Micro

Ibm Granite

5.10

Input

$0.017

Context

131k

G

Google: Gemma 3 27B

Google

4.90

Input

$0.08

Context

131k

ML

Meta: Llama 3.2 1B Instruct

Meta Llama

4.80

Input

$0.027

Context

60k

Methodology

Read the signal, then read the receipt.

Modex keeps each benchmark dimension separate. Intelligence, coding, agentic and reasoning values are not combined into an invented score.

What is shown

The selected value is the score reported by the upstream benchmark dataset and is plotted against catalog input pricing. Prices are normalized per million input tokens and shown in your selected currency.

What it cannot prove

Scores depend on task design, evaluation date, model version and coverage. They do not predict every workload, latency, reliability or total market adoption.

Source and freshness

Catalog and benchmark records are fetched from the connected public API and cached server-side. The timestamp below is Modex fetch time; it is not a claim about the benchmark’s publication date.

Last fetched: Oct 24, 2026, 1:08 PM