AI models for coding: 17 that hold a repository in one pass

Prices from official vendor price lists · October 2026
17 models trained specifically for writing, reviewing and refactoring code across a repository. Output prices run from $0.12 to $14 per 1M tokens across the 14 models with a paid per-token price; 6 have a free tier and 4 ship open weights.
Coding
Services found: 17
#AI modelCapabilitiesAPI $/1M (in / out)Context windowPrice / moOutput $/1MOverviewVisit site
1

Kimi K2.7 Code

Moonshot AI
Text & chatVisionCoding
$0.71 / $3.5262.1K tokensAPI only3.5View
2

Qwen3 Coder Next

Qwen
Text & chatCodingReasoning
$0.12 / $0.8262.1K tokensAPI only0.8View
3

Devstral 2 2512

Mistral AI
Text & chatCodingStructured JSON output
$0.4 / $2262.1K tokensAPI only2View
4

GPT-5.3-Codex

OpenAI
Text & chatVisionCoding
$1.75 / $14400K tokensAPI only14View
5

Codestral

Mistral AI
Text & chatCodingStructured JSON output
$0.3 / $0.9128K tokensAPI only0.9View
6

Devstral 2

Mistral AI
Text & chatCodingStructured JSON output
256K tokensAPI onlyView
7

North Mini Code

Cohere
Text & chatCodingLow latency
256K tokensAPI onlyView
8

Doubao Seed 2.0 Code

ByteDance
Text & chatVisionCoding
API onlyView
9

Laguna S 2.1

Poolside
Text & chatCodingLong context 1M+
$0.09 / $0.181.05M tokensAPI only0.18View
10

Laguna XS 2.1

Poolside
Text & chatCodingMultilingual
$0.06 / $0.12262.1K tokensAPI only0.12View
11

Qwen3 Coder Flash

Qwen
Text & chatVisionCoding
$0.195 / $0.9751M tokensAPI only0.975View
12

GPT-5.2-Codex

OpenAI
Text & chatVisionCoding
$1.75 / $14400K tokensAPI only14View
13

Qwen3 Coder Plus

Qwen
Text & chatCodingReasoning
$0.65 / $3.251M tokensAPI only3.25View
14

GPT-5.1-Codex-Mini

OpenAI
Text & chatVisionCoding
$0.25 / $2400K tokensAPI only2View
15

GPT-5.1-Codex

OpenAI
Text & chatVisionCoding
$1.25 / $10400K tokensAPI only10View
16

GPT-5.1-Codex-Max

OpenAI
Text & chatVisionCoding
$1.25 / $10400K tokensAPI only10View
17

Seed-2.0-Code

Bytedance Seed
Text & chatVisionVideo
$0.5 / $3262.1K tokensAPI only3View

Browse by need

Reasoning models169 modelsCoding assistants17 modelsVision and images156 modelsLong context 1M+82 modelsFree tier available80 modelsOpen weights88 models

Seventeen models here are trained specifically for code: writing it, reviewing it, refactoring across a repository and filling in the middle of an existing file. Kimi K2.7 Code, Qwen3 Coder Next and Devstral 2 lead the dedicated list, with output from free tiers up to $14 per million tokens.

Dedicated coding models are usually cheaper than general flagships and competitive on routine work - boilerplate, tests, migrations, straightforward bug fixes. For large refactors spanning many files the frontier general models still pull ahead; Claude Opus 5 at $25 is the common choice. Check the context window: repository-scale work wants 1M tokens.

Fill-in-the-middleEditing inside a file, not just completing the end — essential for editor integration.
Cheaper than flagshipsDedicated coding models undercut general frontier models on routine work.
Context decidesRepository-scale tasks need a 1M window; snippet work does not.

Models with this capability

The 17 models below all support this capability. Prices are input and output per 1M tokens from the vendor official price list; for dirhams multiply by 3.6725. Tap a card to open the model page.

Kimi K2.7 Code
#1
Moonshot AI
in $0.71out $3.5262.1K tokens
Qwen3 Coder Next
#2
Alibaba (Qwen)
in $0.12out $0.8262.1K tokens
Devstral 2 2512
#3
Mistral AI
in $0.4out $2262.1K tokens
GPT-5.3-Codex
#4
OpenAI
in $1.75out $14400K tokens
Codestral
#5
Mistral AI
in $0.3out $0.9128K tokens
Devstral 2
#6
Mistral AI
price on request256K tokens
North Mini Code
#7
Cohere
in $0out $0256K tokens
Doubao Seed 2.0 Code
#8
ByteDance
price on request

How to choose

  • Match the model to the task size: snippets, single files or whole repositories.
  • Check fill-in-the-middle support if you integrate with an editor.
  • Test on your own stack — training mixes differ sharply by language.
  • Compare against a frontier general model on your hardest refactor before deciding.

Common mistakes

Judging on toy problemsEvery model writes fizzbuzz. Evaluate on your actual codebase.
Ignoring reviewGenerated code compiles more often than it is correct.

Who this suits

Engineering teamsMigrations, test generation and boilerplate at volume.
Solo developersAn always-available pair for unfamiliar languages.
Platform teamsAutomated fixes across many repositories.

Related filters

Code execution47 models that run code in a sandbox
Frontier tier41 flagship models, the best each vendor sells
Open weights88 models you can download and self-host
Long context 1M+82 models holding 1M tokens or more
All 239 modelsThe full rating with filters and prices

Frequently asked questions

Are dedicated coding models better than flagships?

Cheaper and competitive on routine work. For large cross-file refactors the frontier general models still lead.

What is fill-in-the-middle?

Completing code inside an existing file rather than only at the end — essential for editor integrations.

How much context do I need?

Repository-scale work wants 1M tokens. Several options here provide it.

Do coding models handle architecture questions?

Less well than frontier general models. They are optimised for writing, not deciding.

Which is cheapest for bulk work?

Open-weight coder models start near free hosted rates and can be self-hosted.

Frequently searched

By capabilityReasoningCodingVisionLong context 1M+Free tierOpen weightsLow latencyRealtime voiceImage generation
By vendorAnthropic (13)OpenAI (30)Google DeepMind (17)xAI (8)Meta (6)DeepSeek (9)Mistral AI (11)Alibaba (Qwen) (34)Z.ai (12)Moonshot AI (5)
Popular modelsClaude Fable 5.1GPT-6 AstraClaude Opus 5Claude Fable 5GPT-5.6 SolGLM 5.3Grok 4.6Kimi K3GPT-5.6 Terra
By deliveryBatch APIPrompt cachingPriority tierFrontier tierProprietary APITool callingStructured JSONWeb searchMultilingual

Ratings 0

Avg. score 0

Rate this page

Help us improve