Code Models

Data as of June 12, 2026

Model										SWE-ProB	SWEB	Term2B	SWE-MultiB	ArenaB		Notes▼
1.□	Claude Opus 4.8Anthropic	2026/5	?	200K	$15/$75	-	-	📝👁️	○	165.0	188.1	469.2	-	11580	1	SOC2; HIPAA eligible. System card 2026-05-29: 3.7% code-summary dishonesty, 5x fewer dishonest agentic reports vs 4.7; incremental capability gain
2.□	Claude Opus 4.7Anthropic	2026/4	?	200K	$15/$75	-	-	📝👁️	○	264.3	287.6	568.5	-	21573	2	SOC2; HIPAA eligible
3.□	Claude Opus 4.6 ThinkingAnthropic	2026/2	?	200K	$5/$25	40	1.78	📝👁️	○	953.4	480.8	965.4	177.8	31548	3.5	SOC2; HIPAA eligible
4.□	GPT-5.5OpenAI	2026/4	?	400K	-	-	-	📝👁️🔊	○	558.6	382.6	273.2	-	61509	4.5
5.□	Gemini 3.1 ProGoogle	2026/2	?	1M	$2/$12	109	29.7	📝👁️🔊	○	854.2	580.6	668.5	276.9	81456	6.5	-
6.□	Claude Sonnet 4.6Anthropic	2026/2	?	200K	$3/$15	44	1.48	📝👁️	○	-	979.6	-	-	51521	7	SOC2; HIPAA eligible
7.□	Qwen3.7 MaxAlibaba	2026/5	?	1M	$3/$8	-	-	📝👁️	○	360.6	680.4	369.7	-	-	8.5	Qwen3.7 flagship; agent-centric coding. SWE-Bench Verified 80.4, SWE-Bench Pro 60.6.
8.□	GLM 5.1Z.ai	2026/4	754B MoE	200K	$0.95/$3	74	1.64	📝👁️	●	-	1377.8	-	-	41530	9.5	Open-weight; #1 SWE-Bench Pro (58.4)
9.□	Qwen 3.6 PlusAlibaba	2026/4	?	128K	-	96	2.44	📝👁️	○	-	1078.8	-	-	91453	9.5	-
10.□	Kimi K2.6Moonshot	2026/4	1T (32B active, MoE)	256K	-	-	-	📝👁️	●	658.6	780.2	766.7	376.7	-	9.5	Open-weight; Kimi table reports SOTA open coding/agent scores
11.□	MiniMax M2.7MiniMax	2026/3	230B (10B active, MoE)	128K	-	45	2.53	📝👁️	●	-	880.2	-	-	131425	10	-
12.□	Gemini 3 FlashGoogle	2025/12	?	1M	$0.50/$3	159	1.19	📝👁️🔊	○	-	1178.0	-	-	101436	10.5	-
13.□	MIMO v2 ProXiaomi	-	32B	128K	-	129	2.14	📝	●	-	1278.0	-	-	111433	11.5	-
14.□	GLM 4.6Z.ai	2025/7	754B MoE	200K	-	-	-	📝	●	-	1477.8	-	-	-	12	Open-weight MoE; ~216GB VRAM (Q4_K_M); basis for Big Pickle on OpenCode Zen
15.□	GPT 5.4OpenAI	2026/3	?	128K	$2/$14	74	151.8	📝👁️🔊	○	757.7	1776.9	1065.4	-	71457	12	-
16.□	GLM 5V TurboZ.ai	2026/4	744B (40B active, MoE)	200K	$1/$4	-	-	📝👁️🎥	○	-	-	-	-	-	12.5	Native multimodal coding model; CogViT vision encoder; 30+ task joint RL; leads AndroidWorld, WebVoyager, ZClawBench; design-to-code specialist
17.□	Nemotron 3 SuperNVIDIA	2026/4	120B (12B active, MoE)	1M	-	-	-	📝	●	-	-	-	-	-	12.5	Hybrid Mamba-Attention MoE w/ LatentMoE; trained on 25T tokens; 2.2x throughput vs GPT-OSS-120B, 7.5x vs Qwen3.5-122B; native MTP speculative decoding
18.□	Owl AlphaOpenRouter	2026/4	?	1M	free/free	-	-	📝	○	-	-	-	-	-	12.5	High-performance foundation model for agentic workloads; tool use, code generation, automated workflows; compatible with Claude Code and OpenClaw. Provider may log prompts/completions.
19.□	Qwen3.6-27BAlibaba	2026/4	27B	262K	-	-	-	📝	●	-	-	-	-	-	12.5	Open-weight dense; ~80 tps on RTX 5090 with 218K context (vLLM 0.19)
20.□	Qwen3.6-35B-A3BAlibaba	2026/4	35B (3B active)	256K	-	-	-	📝	●	-	-	-	-	-	12.5	Open-weight MoE; 35B total, 3B active; outperforms Claude Opus 4.7 on Simon Willison's pelican test
21.□	Big PickleOpenCode Zen	2025/10	754B MoE	200K	free/free	-	-	📝	○	-	1577.8	-	-	-	12.5	Stealth model (GLM 4.6); free on OpenCode Zen; reasoning + tool calling; text-only
22.□	Llama 3.3 70B (Groq)Groq	2024/12	70B	128K	$0.59/$0.79	394	0.3	📝	●	-	-	-	-	-	12.5	-
23.□	DeepSeek V4DeepSeek	2026/4	1.6T (49B active, MoE)	1M	-	-	-	📝	●	-	-	-	-	-	12.5	China API; self-host OK
24.□	Ring-2.6-1TInclusionAI	2026/5	1T (63B active, MoE)	262K	-	-	-	📝	●	-	-	-	-	-	12.5	Trillion-parameter thinking model; PinchBench 87.6; optimized for coding agents, tool use, multi-turn agent workflows
25.□	Ling-2.6-1TInclusionAI	2026/4	1T	262K	$0.30/$3	-	-	📝	●	-	-	-	-	-	12.5	Instant instruct model; fast-thinking; SOTA on AIME26 and SWE-bench Verified; hybrid MLA + Linear Attention architecture
26.□	Ling-2.6-flashInclusionAI	2026/4	104B (7.4B active, MoE)	262K	-	-	-	📝	●	-	-	-	-	-	12.5	Fast instruct model; lightweight companion to Ling-2.6-1T; coding, document processing, agent workflows
27.□	Llama 4 MaverickMeta	2025/4	-	-	-	-	-	📝👁️	●	-	-	-	-	-	12.5	Meta's flagship coding-capable model.
28.□	MiniMax M3MiniMax	2026/6	?	1M	$0.60/$2	-	-	📝👁️	●	459.0	-	866.0	-	-	12.5	Surpasses GPT-5.5 & Gemini 3.1 Pro on SWE-Bench Pro at ~5-10% of cost; approaches Claude Opus 4.7. SWE-fficiency 34.8.
29.□	Gemini 3.5 FlashGoogle	2026/5	?	1M	$2/$9	-	-	📝👁️🔊	○	-	-	176.2	-	-	12.5	Beats Gemini 3.1 Pro on coding/agentic suite; MCP Atlas 83.6.
30.□	Qwen3.7 PlusAlibaba	2026/6	?	1M	$0.40/$2	-	-	📝👁️🎥	○	-	-	-	-	-	12.5	AA coding 46.5; ScreenSpot Pro 79.0 (GUI grounding).
31.□	Nemotron 3 Ultra 550BNVIDIA	2026/6	550B (55B active, MoE)	1M	-	146.3	-	📝	●	-	-	-	-	-	12.5	43 programming languages; AA Intelligence Index 48.
32.□	MAI-Thinking-1Microsoft AI	2026/6	35B active (SMoE)	256K	-	-	-	📝	○	1053.0	-	-	-	-	12.5	Microsoft's first reasoning model; SWE-Bench Pro 53.
33.□	MAI-Code-1-FlashMicrosoft AI	2026/6	5B active	?	-	-	-	📝	○	1151.0	-	-	-	-	12.5	Inference-efficient agentic coding model for GitHub Copilot/VS Code; Haiku-class size, cheaper.
34.□	Mistral Medium 3.5Mistral	2026/4	128B	256K	$2/$8	-	-	📝👁️	●	-	1677.6	-	-	-	13.5	Merged instruct/reasoning/coding flagship; replaces Devstral 2 in Mistral Vibe; self-hostable on as few as 4 GPUs
35.□	Kimi K2.5 ThinkingMoonshot	2026/1	1T	128K	-	45	2.38	📝👁️	○	1250.7	1876.8	1150.8	473.0	121429	15	China API
36.□	Hy3 PreviewTencent Hunyuan	2026/4	295B (21B active, MoE)	256K	-	-	-	📝	●	-	2174.4	-	-	-	16	Open-weight Hunyuan 3 preview; rebuilt RL/pretraining stack; strong code and agent benchmarks
37.□	Llama 4 Scout (Groq)Groq	2025/4	17Bx16E	128K	$0.11/$0.34	594	0.3	📝👁️	●	-	2274.4	-	-	-	16.5	-
38.□	Qwen 3.5 397B (Together)Together AI	2026/2	397B	128K	$0.27/$0.85	96	-	📝👁️	●	-	1976.4	-	-	151386	17	-
39.□	Mistral Large 3Mistral	2025/12	675B MoE (41B active)	256K	$0.50/$2	48	1.04	📝👁️	●	-	-	-	-	211222	17.5	-
40.□	GPT 5.1OpenAI	2025/11	?	128K	$2/$14	111	57.5	📝👁️🔊	○	-	2076.3	-	-	181339	19	-
41.□	DeepSeek V3.2 ThinkingDeepSeek	2025/9	685B	128K	$0.55/$2	35	3.75	📝	●	-	2473.1	-	-	161368	20	China API; self-host OK
42.□	Grok 4.20 ReasoningxAI	-	?	128K	$0.20/$0.50	233	10.33	📝👁️	○	-	2770.8	-	-	141393	20.5	-
43.□	Claude Haiku 4.5Anthropic	2025/10	?	200K	$1/$5	138	1.16	📝👁️	○	-	2373.3	-	-	201315	21.5	-
44.□	Qwen 3.5 27B (Groq)Groq	2026/2	27B	128K	$0.29/$0.59	662	0.3	📝	●	-	2672.4	-	-	171344	21.5	-
45.□	DeepSeek V3.2 (Together)Together AI	2025/9	685B	128K	$0.28/$0.42	-	-	📝	●	-	2573.1	-	-	191330	22	-
46.□	Devstral 2Mistral	-	24B	128K	-	-	-	📝	●	-	2861.6	-	-	221197	25	-

Claude Opus 4.8

Anthropic?200K$15/$75📝👁️

SWE-Pro 65.0SWE 88.1Term2 69.2Arena 1580

SOC2; HIPAA eligible. System card 2026-05-29: 3.7% code-summary dishonesty, 5x fewer dishonest agentic reports vs 4.7; incremental capability gain

Claude Opus 4.7

Anthropic?200K$15/$75📝👁️

SWE-Pro 64.3SWE 87.6Term2 68.5Arena 1573

SOC2; HIPAA eligible

Claude Opus 4.6 Thinking

#3.5

Anthropic?200K$5/$25401.78📝👁️

SWE-Pro 53.4SWE 80.8Term2 65.4SWE-Multi 77.8Arena 1548

SOC2; HIPAA eligible

GPT-5.5

#4.5

OpenAI?400K📝👁️🔊

SWE-Pro 58.6SWE 82.6Term2 73.2Arena 1509

Gemini 3.1 Pro

#6.5

Google?1M$2/$1210929.7📝👁️🔊

SWE-Pro 54.2SWE 80.6Term2 68.5SWE-Multi 76.9Arena 1456

Claude Sonnet 4.6

Anthropic?200K$3/$15441.48📝👁️

SWE 79.6Arena 1521

SOC2; HIPAA eligible

Qwen3.7 Max

#8.5

Alibaba?1M$3/$8📝👁️

SWE-Pro 60.6SWE 80.4Term2 69.7

Qwen3.7 flagship; agent-centric coding. SWE-Bench Verified 80.4, SWE-Bench Pro 60.6.

GLM 5.1●

#9.5

Z.ai754B MoE200K$0.95/$3741.64📝👁️

SWE 77.8Arena 1530

Open-weight; #1 SWE-Bench Pro (58.4)

Qwen 3.6 Plus

#9.5

Alibaba?128K962.44📝👁️

SWE 78.8Arena 1453

10.

Kimi K2.6●

#9.5

Moonshot1T (32B active, MoE)256K📝👁️

SWE-Pro 58.6SWE 80.2Term2 66.7SWE-Multi 76.7

Open-weight; Kimi table reports SOTA open coding/agent scores

11.

MiniMax M2.7●

#10

MiniMax230B (10B active, MoE)128K452.53📝👁️

SWE 80.2Arena 1425

12.

Gemini 3 Flash

#10.5

Google?1M$0.50/$31591.19📝👁️🔊

SWE 78.0Arena 1436

13.

MIMO v2 Pro●

#11.5

Xiaomi32B128K1292.14📝

SWE 78.0Arena 1433

14.

GLM 4.6●

#12

Z.ai754B MoE200K📝

SWE 77.8

Open-weight MoE; ~216GB VRAM (Q4_K_M); basis for Big Pickle on OpenCode Zen

15.

GPT 5.4

#12

OpenAI?128K$2/$1474151.8📝👁️🔊

SWE-Pro 57.7SWE 76.9Term2 65.4Arena 1457

16.

GLM 5V Turbo

#12.5

Z.ai744B (40B active, MoE)200K$1/$4📝👁️🎥

Native multimodal coding model; CogViT vision encoder; 30+ task joint RL; leads AndroidWorld, WebVoyager, ZClawBench; design-to-code specialist

17.

Nemotron 3 Super●

#12.5

NVIDIA120B (12B active, MoE)1M📝

Hybrid Mamba-Attention MoE w/ LatentMoE; trained on 25T tokens; 2.2x throughput vs GPT-OSS-120B, 7.5x vs Qwen3.5-122B; native MTP speculative decoding

18.

Owl Alpha

#12.5

OpenRouter?1Mfree/free📝

High-performance foundation model for agentic workloads; tool use, code generation, automated workflows; compatible with Claude Code and OpenClaw. Provider may log prompts/completions.

19.

Qwen3.6-27B●

#12.5

Alibaba27B262K📝

Open-weight dense; ~80 tps on RTX 5090 with 218K context (vLLM 0.19)

20.

Qwen3.6-35B-A3B●

#12.5

Alibaba35B (3B active)256K📝

Open-weight MoE; 35B total, 3B active; outperforms Claude Opus 4.7 on Simon Willison's pelican test

21.

Big Pickle

#12.5

OpenCode Zen754B MoE200Kfree/free📝

SWE 77.8

Stealth model (GLM 4.6); free on OpenCode Zen; reasoning + tool calling; text-only

22.

Llama 3.3 70B (Groq)●

#12.5

Groq70B128K$0.59/$0.793940.3📝

23.

DeepSeek V4●

#12.5

DeepSeek1.6T (49B active, MoE)1M📝

China API; self-host OK

24.

Ring-2.6-1T●

#12.5

InclusionAI1T (63B active, MoE)262K📝

Trillion-parameter thinking model; PinchBench 87.6; optimized for coding agents, tool use, multi-turn agent workflows

25.

Ling-2.6-1T●

#12.5

InclusionAI1T262K$0.30/$3📝

Instant instruct model; fast-thinking; SOTA on AIME26 and SWE-bench Verified; hybrid MLA + Linear Attention architecture

26.

Ling-2.6-flash●

#12.5

InclusionAI104B (7.4B active, MoE)262K📝

Fast instruct model; lightweight companion to Ling-2.6-1T; coding, document processing, agent workflows

27.

Llama 4 Maverick●

#12.5

Meta📝👁️

Meta's flagship coding-capable model.

28.

MiniMax M3●

#12.5

MiniMax?1M$0.60/$2📝👁️

SWE-Pro 59.0Term2 66.0

Surpasses GPT-5.5 & Gemini 3.1 Pro on SWE-Bench Pro at ~5-10% of cost; approaches Claude Opus 4.7. SWE-fficiency 34.8.

29.

Gemini 3.5 Flash

#12.5

Google?1M$2/$9📝👁️🔊

Term2 76.2

Beats Gemini 3.1 Pro on coding/agentic suite; MCP Atlas 83.6.

30.

Qwen3.7 Plus

#12.5

Alibaba?1M$0.40/$2📝👁️🎥

AA coding 46.5; ScreenSpot Pro 79.0 (GUI grounding).

31.

Nemotron 3 Ultra 550B●

#12.5

NVIDIA550B (55B active, MoE)1M146.3📝

43 programming languages; AA Intelligence Index 48.

32.

MAI-Thinking-1

#12.5

Microsoft AI35B active (SMoE)256K📝

SWE-Pro 53.0

Microsoft's first reasoning model; SWE-Bench Pro 53.

33.

MAI-Code-1-Flash

#12.5

Microsoft AI5B active?📝

SWE-Pro 51.0

Inference-efficient agentic coding model for GitHub Copilot/VS Code; Haiku-class size, cheaper.

34.

Mistral Medium 3.5●

#13.5

Mistral128B256K$2/$8📝👁️

SWE 77.6

Merged instruct/reasoning/coding flagship; replaces Devstral 2 in Mistral Vibe; self-hostable on as few as 4 GPUs

35.

Kimi K2.5 Thinking

#15

Moonshot1T128K452.38📝👁️

SWE-Pro 50.7SWE 76.8Term2 50.8SWE-Multi 73.0Arena 1429

China API

36.

Hy3 Preview●

#16

Tencent Hunyuan295B (21B active, MoE)256K📝

SWE 74.4

Open-weight Hunyuan 3 preview; rebuilt RL/pretraining stack; strong code and agent benchmarks

37.

Llama 4 Scout (Groq)●

#16.5

Groq17Bx16E128K$0.11/$0.345940.3📝👁️

SWE 74.4

38.

Qwen 3.5 397B (Together)●

#17

Together AI397B128K$0.27/$0.8596📝👁️

SWE 76.4Arena 1386

39.

Mistral Large 3●

#17.5

Mistral675B MoE (41B active)256K$0.50/$2481.04📝👁️

Arena 1222

40.

GPT 5.1

#19

OpenAI?128K$2/$1411157.5📝👁️🔊

SWE 76.3Arena 1339

41.

DeepSeek V3.2 Thinking●

#20

DeepSeek685B128K$0.55/$2353.75📝

SWE 73.1Arena 1368

China API; self-host OK

42.

Grok 4.20 Reasoning

#20.5

xAI?128K$0.20/$0.5023310.33📝👁️

SWE 70.8Arena 1393

43.

Claude Haiku 4.5

#21.5

Anthropic?200K$1/$51381.16📝👁️

SWE 73.3Arena 1315

44.

Qwen 3.5 27B (Groq)●

#21.5

Groq27B128K$0.29/$0.596620.3📝

SWE 72.4Arena 1344

45.

DeepSeek V3.2 (Together)●

#22

Together AI685B128K$0.28/$0.42📝

SWE 73.1Arena 1330

46.

Devstral 2●

#25

Mistral24B128K📝

SWE 61.6Arena 1197

AI App Builders (prompt → working app)

Data as of April 26, 2026

Model										Notes▼
1.□	v0Vercel	2023/10	Custom (built on Claude / GPT)	React / Next.js / Tailwind / shadcn	Vercel functions / API routes	One-click Vercel	Free / $20 Premium / $50 Team	Proprietary	○	First mover in the prompt-to-app space; tightly Vercel-integrated
2.□	Magic PatternsMagic Patterns	2023/12	Claude / GPT	React / Tailwind / shadcn / others	API integrations	GitHub export / Vercel	Free / $30 / $90 Pro+	Proprietary	○	Strong design-fidelity output; screenshot-to-code
3.□	Tempo LabsTempo	2024/1	Claude / GPT	React / Tailwind / shadcn	Supabase / external APIs	Tempo / GitHub export	Free / $30 / $80 Team	Proprietary	○	Visual editor + AI; component-first workflow
4.□	LovableLovable	2024/6	Claude / GPT	React / Tailwind / shadcn	Supabase	Lovable hosting / GitHub export	Free / $20 / $50 / $100 Team	Proprietary	○	Strong Supabase integration; ex-GPT Engineer team
5.□	GitWitGitWit	2024/7	Claude / GPT	React / Vue	GitHub-native	GitHub Actions / direct PR	Free / $20	Proprietary	○	GitHub-first; PR-based collaboration
6.□	WebcrumbsWebcrumbs	2024/8	Claude / GPT	React / Next.js	Webcrumbs / external	Webcrumbs / Vercel / GitHub	Free / $20 / $50	Proprietary	○	Visual canvas; snippet-style component library
7.□	TrickleTrickle	2024/9	Claude / GPT	React	Trickle hosting	Trickle / GitHub	Free / $20 / $50	Proprietary	○	AI-native website builder; landing-page focused
8.□	Replit AgentReplit	2024/9	Claude (managed)	Any (Replit envs)	Replit deployments	Replit hosting (free tier limited)	$25/mo Replit Core	Proprietary	○	Long-tail framework support via Replit envs; agent that drives the IDE
9.□	CodevCodev	2024/9	Claude / GPT	Next.js + Supabase	Supabase	Vercel / GitHub	Free / $25	Proprietary	○	Generates auth-ready full-stack apps end to end
10.□	Bolt.newStackBlitz	2024/10	Claude / GPT / Gemini selectable	React / Vue / Svelte / Astro / others	WebContainers (Node in browser)	Netlify / GitHub / Cloudflare	Free / $20 / $50 Pro+	Proprietary	○	Browser-native; uses WebContainers for full Node runtime
11.□	SameSame	2024/12	Claude / GPT / Gemini	React / Vue / Svelte	Various	Same hosting / Vercel / Netlify	Free / $20 / $40	Proprietary	○	Multi-framework; clone-from-URL feature
12.□	Tldraw ComputerTldraw	2024/12	Claude / GPT	Generated UI in tldraw canvas	n/a (single-page demos)	Browser only	Free	Apache 2.0	●	Draw a layout → generates working interactive UI on the canvas
13.□	Open LovableMendable	2025/8	Any (BYOK)	React / Tailwind	Supabase / others	Self-hosted	Free (self-host)	AGPLv3	●	Open-source clone of Lovable; bring your own model + database

VercelCustom (built on Claude / GPT)React / Next.js / Tailwind / shadcnVercel functions / API routesOne-click VercelFree / $20 Premium / $50 TeamProprietary

First mover in the prompt-to-app space; tightly Vercel-integrated

Magic Patterns

Magic PatternsClaude / GPTReact / Tailwind / shadcn / othersAPI integrationsGitHub export / VercelFree / $30 / $90 Pro+Proprietary

Strong design-fidelity output; screenshot-to-code

Tempo Labs

TempoClaude / GPTReact / Tailwind / shadcnSupabase / external APIsTempo / GitHub exportFree / $30 / $80 TeamProprietary

Visual editor + AI; component-first workflow

Lovable

LovableClaude / GPTReact / Tailwind / shadcnSupabaseLovable hosting / GitHub exportFree / $20 / $50 / $100 TeamProprietary

Strong Supabase integration; ex-GPT Engineer team

GitWit

GitWitClaude / GPTReact / VueGitHub-nativeGitHub Actions / direct PRFree / $20Proprietary

GitHub-first; PR-based collaboration

Webcrumbs

WebcrumbsClaude / GPTReact / Next.jsWebcrumbs / externalWebcrumbs / Vercel / GitHubFree / $20 / $50Proprietary

Visual canvas; snippet-style component library

Trickle

TrickleClaude / GPTReactTrickle hostingTrickle / GitHubFree / $20 / $50Proprietary

AI-native website builder; landing-page focused

Replit Agent

ReplitClaude (managed)Any (Replit envs)Replit deploymentsReplit hosting (free tier limited)$25/mo Replit CoreProprietary

Long-tail framework support via Replit envs; agent that drives the IDE

Codev

CodevClaude / GPTNext.js + SupabaseSupabaseVercel / GitHubFree / $25Proprietary

Generates auth-ready full-stack apps end to end

10.

Bolt.new

StackBlitzClaude / GPT / Gemini selectableReact / Vue / Svelte / Astro / othersWebContainers (Node in browser)Netlify / GitHub / CloudflareFree / $20 / $50 Pro+Proprietary

Browser-native; uses WebContainers for full Node runtime

11.

Same

SameClaude / GPT / GeminiReact / Vue / SvelteVariousSame hosting / Vercel / NetlifyFree / $20 / $40Proprietary

Multi-framework; clone-from-URL feature

12.

Tldraw Computer●

TldrawClaude / GPTGenerated UI in tldraw canvasn/a (single-page demos)Browser onlyFreeApache 2.0

Draw a layout → generates working interactive UI on the canvas

13.

Open Lovable●

MendableAny (BYOK)React / TailwindSupabase / othersSelf-hostedFree (self-host)AGPLv3

Open-source clone of Lovable; bring your own model + database