Speech, image and video models

New speech, image, video and world models, reported by their developers and listed in catalogues. Recorded automatically.

Thu 5 Nov

GoogleGoogle releases Gemini Omni 1.2 FlashUpdated2 evidence links

Context window: 131K
Input · USD / 1M tokens: $1.50 (text / image / video)
Output · USD / 1M tokens: $9.00 (text); $12.00 ((video)*)

4 Nov 23:15spottedVertex AI
5 Nov 16:00spottedVertex AI
5 Nov 17:08spottedGoogle DeepMind

Google DeepMind · blog Google · AI blog Evidence links Model page

Wed 28 Oct

NVIDIANemotron 3.5 Transcribe 30B A4B: Open weightsOpen weights1 evidence link

Context window: 66K

Hugging Face · blog Evidence links Model page

Tue 27 Oct

Liquid AILFM3 Omni 2B: Open weightsOpen weights2 evidence links

Hugging Face · trending Primary source (reviewed)Evidence links Model page

GoogleGenie 4: Research previewResearch preview1 evidence link

Google DeepMind Evidence links Model page

MicrosoftMAI-Image-2.7: Public previewPublic preview2 evidence links

Arena Primary source (reviewed)Evidence links Model page

Mon 26 Oct

TencentHy-SimulMT-30B-A3B-Interleave: Open weightsOpen weights4 evidence links

Hugging Face · tencent Hugging Face · tencent Hugging Face · tencent Evidence links Model page

Tue 20 Oct

MistralVoxtral Small Transcribe 3: Catalogue listingUpdated5 evidence links

Context window: 66K

Mistral API Vercel AI Gateway models.dev · catalogue Evidence links Model page

MistralVoxtral Mini Transcribe 3: Catalogue listingUpdated5 evidence links

Context window: 66K

Mistral API Vercel AI Gateway models.dev · catalogue Evidence links Model page

Thu 15 Oct

OpenAIGPT-Live-1.1: Product updateProduct update1 evidence link

OpenAI · ChatGPT release notes Evidence links Model page

MetaMuse Spark Open: Open weightsUpdated12 evidence links

Context window: 1M

Hugging Face · meta-models Hugging Face · meta-models OpenRouter Evidence links Model page

Black Forest LabsFLUX 3 Image Klein: Open weightsUpdated5 evidence links

Hugging Face · blog Arena Hacker News · front page Evidence links Model page

Wed 14 Oct

xAIGrok Voice 3.0: API availabilityAPI availability5 evidence links

xAI · release notes xAI models.dev · catalogue Evidence links Model page

Mon 12 Oct

MetaMuse Wonder: API availabilityAPI availability13 evidence links

Context window: 1M
Input · USD / 1M tokens: $4.00
Output · USD / 1M tokens: $20.00

12 OctAPI availabilityOpenCode
12 Oct 18:02API availabilityOpenCode
10 Nov 19:23API availabilityOpenCode

Alex's take

Alex · 12 Oct“Meta just released Muse Wonder According to the benchmarks, it's right up there with GPT 6 Astra and Claude Opus 5.5. The results look impressive, but the model might be heavily optimized for benchmarks. It's still…”

OpenCode · discovery models.dev · catalogue OpenRouter Evidence links Model page

Tue 6 Oct

GoogleNano Banana 2.1: API availabilityAPI availability13 evidence links

Context window: 131K
Input · USD / 1M tokens: $1.50 (text/image/video)
Output · USD / 1M tokens: $30.00 (images); $7.50 (text and thinking)

Gemini API Gemini · deprecations Vercel AI Gateway Evidence links Model page

Mon 5 Oct

ElevenLabsEleven v4 Turbo: API availabilityInput / output price3 evidence links

Artificial Analysis · text-to-speech arena Primary source (reviewed)Primary source (reviewed)Evidence links Model page

Thu 1 Oct

MicrosoftMAI-Transcribe-2-Streaming: Public previewUpdated3 evidence links

models.dev · catalogue TrueFoundry · Azure catalogue Primary source (reviewed)Evidence links Model page

MicrosoftMAI-Voice-2.1: Public previewPublic preview3 evidence links

Vercel AI Gateway models.dev · catalogue Primary source (reviewed)Evidence links Model page

MicrosoftMAI-Voice-2.1-Flash: Public previewPublic preview3 evidence links

Vercel AI Gateway models.dev · catalogue Primary source (reviewed)Evidence links Model page

Tue 22 Sep

GoogleGoogle releases Gemini 3.8 Flash TTSInput / output price7 evidence links

Context window: 8K
Input · USD / 1M tokens: $0.50 (text)
Output · USD / 1M tokens: $9.00 (audio)

22 Sep 16:00spottedGemini API
23 Sep 15:48spottedGemini API

Gemini API Gemini · deprecations Google AI for Developers Evidence links Model page

GoogleGoogle releases Gemini 3.8 Flash Lite TTSInput / output price7 evidence links

Context window: 8K
Input · USD / 1M tokens: $0.50 (text)
Output · USD / 1M tokens: $6.00 (audio)

Gemini API Gemini · deprecations Google AI for Developers Evidence links Model page

Fri 18 Sep

xAIIntroducing Grok Voice Transcribe 2.0 spotted on xAIUpdated4 evidence links

xAI · release notes xAI models.dev · catalogue Evidence links Model page