Cerebras has been demonstrating its ability to host large MoEs at very high speeds this week, launching Qwen3 235B 2507 and Qwen3 Coder 480B endpoints at >1,500 output tokens/s ➤ @CerebrasSystems now offers endpoints for both Qwen3 235B 2507 Reasoning & Non-reasoning. Both models have 235B total parameters with 22B active. ➤ Qwen 3 235B 2507 Reasoning offers intelligence comparable to o4-mini (high) & DeepSeek R1 0528. The Non-reasoning variant offers intelligence comparable to Kimi K2 and well above GPT-4.1 and Llama 4 Maverick. ➤ Qwen3 Coder 480B has 480B total parameters with 35B active. This model is particularly strong for agentic coding and can be used in a variety of coding agent tools, including the Qwen3-Coder CLI. Cerebras’ launches represent the first time this level of intelligence has been accessible at these output speeds and have the potential to unlock new use cases - like using a reasoning model for each step of an agent without having to wait minutes.

🇰🇷 LG recently launched EXAONE 4.0 32B - it scores 62 on Artificial Analysis Intelligence Index, the highest score for a 32B model yet @LG_AI_Research's EXAONE 4.0 is released in two variants: the 32B hybrid reasoning model we’re reporting benchmarking results for here, and a smaller 1.2B model designed for on-device applications that we have not benchmarked yet. Alongside Upstage's recent Solar Pro 2 release, it's exciting to see Korean AI labs join the US and China near the top of the intelligence charts. Key results: ➤ 🧠 EXAONE 4.0 32B (Reasoning): In reasoning mode, EXAONE 4.0 scores 62 on the Artificial Analysis Intelligence Index. This matches Claude 4 Opus and the new Llama Nemotron Super 49B v1.5 from NVIDIA, and sits only 1 point behind Gemini 2.5 Flash ➤ ⚡ EXAONE 4.0 32B (Non-Reasoning): In non-reasoning mode, EXAONE 4.0 scores 51 on the Artificial Analysis Intelligence Index. It matches Llama 4 Maverick in intelligence despite having only ~1/4th total parameters (although has ~2x the active parameters) ➤ ⚙️ Output tokens and verbosity: In reasoning mode, EXAONE 4.0 used 100M output tokens for the Artificial Analysis Intelligence Index. This is higher than some other frontier models, but aligns with recent trends of reasoning models using more output tokens to 'think more' - similar to Llama Nemotron Super 49B v1.5, Grok 4, and Qwen3 235B 2507 Reasoning. In non-reasoning mode, EXAONE 4.0 used 15M tokens - high for a non-reasoner, but not as high as Kimi K2’s 30M. Key details: ➤ Hybrid reasoning: The model offers optionality between 'reasoning' mode and 'non-reasoning' mode ➤ Availability: Hosted by @friendliai currently, and competitively priced (especially compared to proprietary options) by FriendliAI at $1 per 1M input and output tokens ➤ Open weights: EXAONE 4.0 is an open weights model available under the EXAONE AI Model License Agreement 1.2. The license limits commercial use. ➤ Multimodality: Text only input and output ➤ Context window: 131k tokens ➤ Parameters: 32B active and total parameters, available in 16bit and 8bit precision (means the model can be run on a single H100 chip in full precision)

Announcing the Artificial Analysis Music Arena Leaderboard: with >5k votes, Suno v4.5 is the leading Music Generation model followed by Riffusion’s FUZZ-1.1 Pro. Google’s Lyria 2 places third in our Instrumental leaderboard, and Udio’s v1.5 Allegro places third in our Vocals leaderboard. The Instrumental Leaderboard is as follows: 🥇 @SunoMusic V4.5 🥈 @riffusionai FUZZ-1.1 Pro 🥉 @GoogleDeepMind Lyria 2 @udiomusic v1.5 Allegro @StabilityAI Stable Audio 2.0 @metaai MusicGen Rankings are based on community votes across a diverse range of genres and prompts. Want to see your prompt featured? You can submit prompts in the arena today. 👇 See below for the Vocals Leaderboard and link to participate!

Model demand change 2024 to 2025: Google (+49pts), DeepSeek (+53pts) and xAI (+31pts) have achieved massive gains in demand share over the past year @Google has transitioned from being an AI laggard to an AI leader with a ~2.5x increase in proportion of respondents using or considering the Gemini model series. A key driver of this has been Google making significant gains in intelligence: Gemini 2.5 Pro now sits at #3 in our Artificial Analysis Intelligence Index, compared to significantly lagging behind OpenAI and Anthropic in early 2024. @deepseek_ai in H1 2024 had only released DeepSeek 67B, a model that saw limited adoption and underperformed Llama 3 70B. DeepSeek first saw some uptake in late 2024 with the releases of their V2 model, and then saw rapid adoption in early 2025 with their V3 and R1 models that have taken them to leadership among open weights models. @xai released its first model Grok-1 in mid-H1 2024 and has since rapidly climbed to intelligence leadership across all models with successive releases, culminating in last week's launch of Grok 4. Source: Artificial Analysis AI Adoption Survey H1 2025 (report available on the Artificial Analysis website)

Kimi K2 Providers: Groq is serving Kimi K2 at >400 output tokens/s, 40X faster than Moonshot’s first-party API Congratulations to a number of providers to being quick to launch APIs for Kimi K2, including @GroqInc , @basetenco , @togethercompute, @FireworksAI_HQ, @parasail_io, @novita_labs, @DeepInfra, and of course @Kimi_Moonshot. This is impressive considering the size of the model at 1 trillion total parameters. Groq stands out for blazing fast speed. DeepInfra, Novita and Baseten stand out for their pricing, being the only providers pricing similarly to or more cheaply than Moonshot’s first party API. See below for further comparisons between the providers. We’re expecting fast increases in speed across some providers as teams optimize for the K2 model - our numbers below show median speeds over the last 72 hours but we’re already seeing DeepInfra jump up to 62 tokens/s in today’s measurements

While Moonshot AI’s Kimi k2 is the leading open weights non-reasoning model in the Artificial Analysis Intelligence Index, it outputs ~3x more tokens than other non-reasoning models, blurring the lines between reasoning & non-reasoning Kimi k2 is the largest major open weights model yet - 1T total parameters with 32B active (this requires a massive 1TB of memory at native FP8 to hold the weights). We have k2 at 57 in Artificial Analysis Intelligence Index, an impressive score that puts it above models like GPT-4.1 and DeepSeek V3, but behind leading reasoning models. Until now, there has been clear a distinction between reasoning model and non-reasoning models in our evals - defined not only by whether the model uses <reasoning> tags, but primarily by token usage. The median number of tokens used to answer all the evals in Artificial Analysis Intelligence Index is ~10x higher for reasoning models than for non-reasoning models. @Kimi_Moonshot's Kimi k2 uses ~3x the number of tokens that the median non-reasoning model uses. Its token usage is only up to 30% lower than Claude 4 Sonnet and Opus when run in their maximum budget extended thinking mode, and is nearly triple the token usage of both Claude 4 Sonnet and Opus with reasoning turned off. We therefore recommend that Kimi k2 be compared to Claude 4 Sonnet and Opus in their maximum budget extended thinking modes, not to the non-reasoning scores for the Claude 4 models. Kimi k2 is available on @Kimi_Moonshot’s first-party API as well as @FireworksAI_HQ, @togethercompute, @novita_labs, and @parasail_io. See below and on Artificial Analysis for further analysis 👇

Less than 500 days since releasing their first model in March 2024, xAI has released the leading AI model

OpenAI's new Deep Research API costs up to ~$30 per API call! These new Deep Research API endpoints might just be the new fastest way to spend money Across our 10 deep research test queries, we spent $100 on o3 and $9.18 on o4-mini. How do the costs get so big? High prices and millions of tokens. These endpoints are versions of o3 and o4-mini that have been RL’d for deep research tasks. Availability via API allows them to be used with both OpenAI’s web search tool and custom data sources via remote MCP servers. o4-mini-deep-research pricing is 5x lower than o3-deep-research pricing. In our test queries, o4-mini also seems to use fewer tokens - it is came in over 10x cheaper in total across our 10 test queries. Pricing: ➤ o3-deep-research is priced at $10 /M input ($2.50 cached input), $40 /M output ➤ o4-mini-deep-research is priced at $2 /M input ($0.5 cached input), $8 /M output These endpoints are both substantially more expensive than OpenAI’s standard o3 and o4-mini endpoints - those are at: ➤ o3: $2 /M ($0.5 cached) input, $8 /M output for o3 ➤ o4-mini: $1.1 /M (0.275 cached) input, $4.4 /M output

Black Forest Labs is going to create a flood of new startups with their open weights image editing model released today - Virtual try-on will get 10X better but that is only the start. We will also see new experiences as people get creative with these models (much more expansive than Snapchat & Instagram filters) - The model is only 12B and can be fine-tuned on consumer hardware - Platforms like @FAL offer full LoRA training support Credit to @FAL for the the image below, they have a great write-up of their fine-tuning offering (link below)

Image editing is now open source! Black Forest Labs just released an open weights image editing model comparable in performance to proprietary models @bfl_ml has released FLUX.1 Kontext [dev], a 12B image editing model. We were given pre-release access and have been testing it in our Artificial Analysis Image Arena. We have independently verified that it offers comparable or superior performance to a number of proprietary models including Google's Gemini 2.0 Flash and ByteDance's Bagel. FLUX.1 Kontext [dev] only trails Black Forest Labs' own proprietary models and OpenAI's GPT-4o. Hosted APIs are also available on @FAL , @replicate, @togethercompute Link below to the weights on HuggingFace 👐