USA · 51 models
The American giants keep the edge on the proprietary frontier — GPT-5.5, Claude Opus 4.7, Gemini 3.1 Pro — but the open-weight ecosystem is thinner (Llama, Phi, Nemotron). Worth a close look if you want to understand the frontier.
| Provider | Model | License | What it's for | Price (input / output) | Context |
|---|---|---|---|---|---|
| OpenAI | GPT-5.5LLM | proprietary | OpenAI's most powerful model. Reach for it when you want the best possible answer and budget takes a back seat. | $5 / $30 /Mtok | 270k |
| GPT-5.4 testedLLM | proprietary | OpenAI's value pick for everyday work: assistant, agent, basic coding. | $2.50 / $15 /Mtok | 270k | |
| GPT-5.4-miniLLM | proprietary | Small, fast model. For high-volume workflows where every cent counts. | $0.75 / $4.50 /Mtok | 270k | |
| GPT-5.3-CodexCode LLM | proprietary | Built for code: it refactors, debugs and writes tests. The right hand of anyone who codes. | Codex Enterprise pricing | 270k | |
| GPT-5.3 InstantFast LLM | proprietary | Answers in under a second, perfect when your chatbot gets a lot of visitors. You trade a bit of quality for speed. | Included in Pro plan | 270k | |
| GPT-Realtime-2Audio | proprietary | Real-time voice conversation with interruption. For phone assistants or voice companions. | $32 / $64 audio /Mtok | — | |
| GPT-Realtime-TranslateAudio | proprietary | Simultaneous real-time translation. Live events, international calls. | $0.034 / min | — | |
| Whisper-RealtimeSTT | proprietary | Real-time speech-to-text. Live transcription, automatic subtitles. | $0.017 / min | — | |
| GPT-Image-2Image | proprietary | Image generation + editing. More accurate on text than Dall-E 3. | $8 / $30 /Mtok img | — | |
| text-embedding-3-largeEmbedding | proprietary | Turns your documents into vectors for RAG. The go-to when you want quality. | $0.13 /Mtok | 8k | |
| text-embedding-3-smallEmbedding | proprietary | Low-cost embedding for high-volume RAG. 6× cheaper than the large one. | $0.02 /Mtok | 8k | |
| Anthropic | Claude Opus 4.7LLM | proprietary | The most subtle model on the market. For long-form writing, nuanced reasoning, high-end prose. | $5 / $25 /Mtok | 200k |
| Claude Opus 4.7 FastLLM | proprietary | Opus 4.7 with latency priority. Pricey, but you get an answer in seconds. | $30 / $150 /Mtok | 200k+ | |
| Claude Opus 4.6 / 4.5LLM | proprietary | Previous Opus versions, still in service. Comparable to 4.7 for 95% of uses. | $5 / $25 /Mtok | 200k | |
| Claude Sonnet 4.6 testedLLM | proprietary | The best balance at Anthropic. My default pick for CRM, agents and pro content. | $3 / $15 /Mtok | 200k (1M β) | |
| Claude Sonnet 4.5LLM | proprietary | The stable Sonnet before 4.6. If you already have a workflow tested on it, keep it. | $3 / $15 /Mtok | 200k (1M β) | |
| Claude Haiku 4.5 testedFast LLM | proprietary | Little Claude for classification, extraction and simple high-volume tasks. | $1 / $5 /Mtok | 200k | |
| Gemini 3.5 FlashMultimodal LLM | proprietary | Google's fastest, with 1M of context. For handling long PDFs or entire codebases. | $1.50 / $9 /Mtok | 1M | |
| Gemini 3.1 ProMultimodal LLM | proprietary | Google's high end. Excellent at native multimodal: image + audio + video in a single call. | $2-4 / $12-18 /Mtok | 1M+ | |
| Gemini 3.1 Flash-LiteMultimodal LLM | proprietary | Mini Gemini, really cheap. For crunching mountains of documents without blowing the budget. | $0.25 / $1.50 /Mtok | 1M | |
| Gemini 3.1 Flash LiveAudio | proprietary | Real-time audio conversation at Google. A direct rival to GPT-Realtime. | $0.75-3 / $4.50-12 /Mtok | 1M | |
| Gemini 3.1 Flash Image PreviewImage | proprietary | Image generation built into the text model. Handy for multimodal prompts. | $0.50 in / 60$/Mimg out | — | |
| Gemini 3 Pro Image PreviewImage | proprietary | High-end image generation. The quality has caught up to Midjourney. | $2 in / 120$/Mimg out | — | |
| Gemini 2.5 ProMultimodal LLM | proprietary | Version 2.5, still heavily used. 2M of context (the largest on the market). | $1.25-2.50 / $10-15 /Mtok | 2M | |
| Gemini 2.5 Flash testedMultimodal LLM | proprietary | Google's cheap Swiss Army knife. Reads text, image, audio and video natively at $0.30/Mtok. | $0.30 / $2.50 /Mtok | 1M | |
| Gemini 2.5 Flash-LiteMultimodal LLM | proprietary | Even smaller than Flash. For batch jobs where volume crushes the budget. | $0.10 / $0.40 /Mtok | 1M | |
| Nano Banana (Gemini 2.5 Flash Image) testedImage | proprietary | The value benchmark for photo editing. It's what I use for Airbnb and personal branding. | $0.039 / image | — | |
| Imagen 4 Fast / Standard / UltraImage | proprietary | Premium photorealistic generation. Ultra goes toe-to-toe with Midjourney v7. | $0.02 / $0.04 / $0.06 per image | — | |
| Veo 3.1 Standard / Fast / LiteVideo | proprietary | Google's direct answer to Sora. Standard goes up to 4K. | $0.05 → $0.60 / s by res. | — | |
| Veo 2Video | proprietary | Previous version, still available. 720p at $0.35/s. | $0.35 / s | — | |
| Lyria 3Audio music | proprietary | Google's music generation. A direct rival to Suno, but built into the Gemini API. | Per request (preview) | — | |
| Gemma 4LLM | open-weight | Google's open models. Free to self-host, strong for their size. | Free weights | — | |
| xAI | Grok 4.3LLM | proprietary | xAI's high-end model (Elon Musk). Strengths: humor, real-time X access, scientific reasoning. | $1.25 / $2.50 /Mtok | 1M |
| Grok Build 0.1Code LLM | proprietary | Grok's code variant. Still in preview. | Preview | 1M | |
| Grok Voice TTSAudio | proprietary | Grok's voice. Character, custom accent, X Spaces integration. | $15 / Mchars TTS | — | |
| Grok Imagine (image)Image | proprietary | xAI image generation. A bit more permissive in style than the others. | $0.02 / image | — | |
| Grok Imagine (video)Video | proprietary | AI video at xAI. Aggressive pricing vs Runway/Pika. | $0.05 / s 720p | — | |
| Meta | Llama 4 MaverickMultimodal LLM | open-weight | Meta's flagship open-weight. 10M context: it swallows the equivalent of a 5,000-page book in a single call. | Free weights · $0.19-0.49 hosted | 10M |
| Llama 4 ScoutMultimodal LLM | open-weight | Smaller Llama 4. Same 10M context, faster to serve. | Free weights | 10M | |
| Llama 3.3 70BLLM | open-weight | The stable classic. Still widely deployed in production. | Free weights | 128k | |
| Llama 3.2 (1B / 3B / 11B / 90B)LLM (+vision on 11B/90B) | open-weight | The full Llama 3.2 range. The 1B and 3B run on a smartphone. | Free weights | 128k | |
| Llama 3.1 (8B / 70B / 405B)LLM | open-weight | The historic tier-1 of the open world. 405B is still the biggest open Llama model. | Free weights | 128k | |
| Microsoft | Phi-4 / mini / multimodalLLM | open-weight | Microsoft's small MIT-licensed models. Excellent at reasoning for their size. | Free weights · Azure | 128k |
| NVIDIA | Nemotron Nano 9B v2 / Super 49BLLM | open-weight | NVIDIA models optimized for their GPUs. Tight pricing via NIM enterprise. | Free weights · $4500/GPU/yr | 128k |
| Databricks | DBRX 132BLLM (MoE) | open-weight | Databricks' open MoE. Above Llama 3 70B on code benchmarks. | $0.75 / $2.25 /Mtok Mosaic | 32k |
| Runway | Gen-4.5 / Gen-4 / Turbo / Gen-3Video | proprietary | Runway's video models. Gen-4.5 = film quality, Turbo = creative speed. | $12/mo (625 credits) | — |
| Aleph / Act-Two / GWM-1Video editing | proprietary | Runway tools for editing an existing video (relight, perf capture, world model). | Credits included | — | |
| Luma | Ray3.14 / Ray3.14 HDRVideo | proprietary | Luma's AI video. Very good at cinematic camera movement. | $30 → $300 / mo | — |
| Pika | Pika 2.5 + 7 variantsVideo | proprietary | The consumer AI video app. Specialties: effects, swaps, transitions. | $8 / $28 / $76 per month | — |
| ElevenLabs (UK/US) | Eleven v3 / Music v1 / SFX v2 / Scribe testedAudio | proprietary | The multilingual TTS benchmark. v3 does multi-speaker, Music composes, SFX generates sound effects. | ~$0.17-0.36 / min | — |
| Suno | Suno v4.5 / v5Audio music | proprietary | Consumer music generation. A full song — lyrics + vocals + instrumentals — from one prompt. | $10 / mo Pro (≈500 tracks) | — |