Back to AI Coding

AI Model Ranking

AI Coding Model Rankings

Snapshot: 2026-07-08

Top 20 models

01

Current reasoning model

Anthropic - Claude Fable - 5

Highest-ranked frontier reasoning entry in the snapshot, with a very large context window.

Best for: Deep repository analysis, complex planning, and high-value agent tasks where latency is acceptable.

Context
1M
AA Index
60
Blended price
$7.70/1M
Speed
71 tok/s
First chunk
149.84s
Total response
156.91s
02

Current reasoning model

Anthropic - Claude Opus - 4.8 (max)

High-intelligence Claude Opus variant with 1M context and stronger response time than Fable.

Best for: Large code migrations, multi-file edits, and coding agents that need strong reasoning.

Context
1M
AA Index
56
Blended price
$3.85/1M
Speed
66 tok/s
First chunk
41.75s
Total response
49.32s
03

Current reasoning model

OpenAI - GPT - 5.5 (xhigh)

Top OpenAI reasoning tier in the snapshot, with high intelligence and broad context.

Best for: OpenAI-centered coding agents, design-to-code work, code review, and tool-heavy workflows.

Context
922k
AA Index
55
Blended price
$4.35/1M
Speed
85 tok/s
First chunk
94.38s
Total response
100.28s
04

Current reasoning model

Anthropic - Claude Opus - 4.7 (max)

Strong Opus reasoning with 1M context and lower latency than the higher Opus/Fable entries.

Best for: Interactive coding sessions that still need frontier reasoning and long context.

Context
1M
AA Index
54
Blended price
$3.85/1M
Speed
56 tok/s
First chunk
19.51s
Total response
28.37s
05

Current reasoning model

Anthropic - Claude Sonnet - 5 (max)

Sonnet-level cost profile with high intelligence and 1M context, trading off long first response.

Best for: Long, careful batch work where quality and context matter more than immediacy.

Context
1M
AA Index
53
Blended price
$1.54/1M
Speed
89 tok/s
First chunk
191.31s
Total response
196.94s
06

Current reasoning model

OpenAI - GPT - 5.5 (high)

A high-reasoning GPT-5.5 mode with the same broad context and much lower latency than xhigh.

Best for: Daily agent coding, code review, refactors, and product engineering loops.

Context
922k
AA Index
53
Blended price
$4.35/1M
Speed
83 tok/s
First chunk
26.89s
Total response
32.90s
07

Current reasoning model

Z AI - GLM - 5.2 (max)

High ranking with 1M context, low listed blended price, and very fast median output.

Best for: Cost-aware coding assistants, long-context analysis, and latency-sensitive workflows.

Context
1M
AA Index
51
Blended price
$0.90/1M
Speed
220 tok/s
First chunk
1.41s
Total response
12.79s
08

Current reasoning model

OpenAI - GPT - 5.5 (medium)

Balanced GPT-5.5 reasoning tier with broad context and faster response than high/xhigh.

Best for: Most interactive coding loops where quality, speed, and context all matter.

Context
922k
AA Index
50
Blended price
$4.35/1M
Speed
73 tok/s
First chunk
11.17s
Total response
18.05s
09

Current general model

Google - Gemini - 3.5 Flash

Fast long-context Gemini model with strong intelligence and Google ecosystem fit.

Best for: Long-file analysis, multimodal coding workflows, and Google Cloud-oriented stacks.

Context
1M
AA Index
50
Blended price
$1.31/1M
Speed
192 tok/s
First chunk
25.21s
Total response
27.81s
10

Current reasoning model

Google - Gemini - 3.1 Pro Preview

Pro preview with 1M context, useful when long-context quality is the main requirement.

Best for: Architecture review, large document-to-code tasks, and multimodal product work.

Context
1M
AA Index
46
Blended price
$1.74/1M
Speed
145 tok/s
First chunk
27.42s
Total response
30.87s
11

Current reasoning model

Alibaba - Qwen - 3.7 Max

High-throughput Qwen entry with 1M context and competitive listed price.

Best for: Multilingual coding, China-friendly model routing, and fast long-context workloads.

Context
1M
AA Index
46
Blended price
$1.43/1M
Speed
205 tok/s
First chunk
2.55s
Total response
16.71s
12

Current reasoning model

Google - Gemini - 3.5 Flash (medium)

A faster Gemini Flash configuration with 1M context and estimated intelligence marker.

Best for: Responsive coding help, long context prompts, and lower-latency Gemini workflows.

Context
1M
AA Index
45*
Blended price
$1.31/1M
Speed
191 tok/s
First chunk
19.51s
Total response
22.12s
13

Current reasoning model

MiniMax - MiniMax - M3

Cost-efficient 1M-context option with strong ranking for its listed price.

Best for: Budget-aware long-context coding, summarization, and batch analysis.

Context
1M
AA Index
44
Blended price
$0.22/1M
Speed
102 tok/s
First chunk
1.87s
Total response
26.31s
14

Current reasoning model

DeepSeek - DeepSeek V4 Pro - max

Very low listed blended price with 1M context and a top-20 intelligence score.

Best for: Cost-sensitive model stacks, alternative coding agents, and long-context experiments.

Context
1M
AA Index
44
Blended price
$0.18/1M
Speed
72 tok/s
First chunk
1.73s
Total response
69.47s
15

Current reasoning model

OpenAI - GPT Codex - 5.3 (xhigh)

Coding-focused OpenAI entry with 400k context and high reasoning configuration.

Best for: Codex-style software generation, repo edits, test fixes, and code review prompts.

Context
400k
AA Index
44*
Blended price
$1.87/1M
Speed
99 tok/s
First chunk
98.73s
Total response
103.78s
16

Current reasoning model

Kimi - Kimi - K2.6

Competitive intelligence with moderate context and low first-token latency.

Best for: Regional coding workflows, prompt iteration, and medium-context agent tasks.

Context
256k
AA Index
44
Blended price
$0.70/1M
Speed
76 tok/s
First chunk
2.37s
Total response
67.37s
17

Current reasoning model

OpenAI - GPT - 5.5 (low)

Low-latency GPT-5.5 tier with broad context and much faster interaction.

Best for: Interactive pair programming, quick fixes, and fast planning loops.

Context
922k
AA Index
43
Blended price
$4.35/1M
Speed
80 tok/s
First chunk
1.79s
Total response
8.07s
18

Current model; partial public metrics

Meta - Muse - Spark

Top-20 intelligence entry with partial public operating metrics in the snapshot.

Best for: Evaluation watchlists and teams comparing Meta model availability.

Context
262k
AA Index
43
Blended price
--/1M
Speed
-- tok/s
First chunk
--
Total response
--
19

Current general model

Anthropic - Claude Opus - 4.7 (Non-reasoning, high)

Fast non-reasoning Claude Opus mode with 1M context and low first-token latency.

Best for: Responsive Claude workflows, code explanation, and long-context edits that do not need heavy reasoning.

Context
1M
AA Index
43*
Blended price
$3.85/1M
Speed
50 tok/s
First chunk
1.52s
Total response
11.42s
20

Current reasoning model

Xiaomi - MiMo - V2.5-Pro

Low listed price with 1M context and a top-20 intelligence placement.

Best for: Cost-focused experimentation, long-context comparisons, and alternative model routing.

Context
1M
AA Index
42
Blended price
$0.18/1M
Speed
52 tok/s
First chunk
2.77s
Total response
50.60s