benchr
The major labs ship something significant every few weeks. This page lists significant releases and material platform changes since early 2026, in reverse chronological order. For deeper coverage of each model, see the linked article or the comparison tool.
Every 2026 release, by date
One row per model, newest first. The date is the day the provider shipped it, the price is its list rate per million tokens, and a dash is a figure the provider has not published — never a value carried over from a related model.
Jump to a month: September 2026 · August 2026 · July 2026 · June 2026 · May 2026 · April 2026 · March 2026 · February 2026
| Released | Model | Provider | Context | $ / 1M in → out |
|---|---|---|---|---|
| Grok Voice Transcribe 2.0 | xAI | — | — | |
| Gemini 3.8 Live Extended Thinking | 128K | — | ||
| Gemini 3.8 Live | 128K | — | ||
| Fugu Ultra v2 | Sakana AI | — | $5.00 → $30.00 | |
| Fugu Max | Sakana AI | — | $2.00 → $6.00 | |
| DeepSeek-V4.1-Flash | DeepSeek | 1M | $0.3 → $1.2 | |
| GPT-6 Astra | OpenAI | 1.05M | $10.00 → $50.00 | |
| Muse Spark 1.3 | Meta | 1M | $1.25 → $4.25 | |
| Gemini 3.8 Flash | 1M | $0.75 → $3.75 | ||
| Claude Mythos 5.1 | Anthropic | 1M | $10.00 → $50.00 | |
| Claude Fable 5.1 | Anthropic | 1M | $10.00 → $50.00 | |
| Qwen3.8-Flash | Alibaba (Qwen) | 256K | $0.15 → $0.47 | |
| GLM-5.3-Flash | Z.ai | 1M | $0.15 → $0.5 | |
| DeepSeek-V4-Flash-Vision-Exp | DeepSeek | 1M | $0.44 → $1.32 | |
| GLM-5.3 | Z.ai | 1M | $1.4 → $4.4 | |
| Gemini 3.7 Flash | 1M | $0.75 → $3.75 | ||
| Grok 4.6 | xAI | 500K | $2.00 → $6.00 | |
| Muse Spark 1.2 | Meta | 1M | $1.25 → $4.25 | |
| Qwen3.8-Max | Alibaba (Qwen) | 1M | $2.00 → $6.00 | |
| Claude Opus 5 | Anthropic | 1M | $5.00 → $25.00 | |
| Gemini 3.6 Flash | 1M | $0.75 → $3.75 | ||
| Gemini 3.5 Flash-Lite | 1M | $0.3 → $2.5 | ||
| Kimi K3 | Moonshot AI | 1M | $3.00 → $15.00 | |
| Qwen-Audio 3.0 TTS Plus | Alibaba (Qwen) | — | — | |
| Qwen-Audio 3.0 TTS Flash | Alibaba (Qwen) | — | — | |
| Muse Spark 1.1 | Meta | 1M | $1.25 → $4.25 | |
| GPT-5.6 Terra | OpenAI | 1.05M | $2.00 → $12.00 | |
| GPT-5.6 Sol | OpenAI | 1.05M | $4.00 → $20.00 | |
| GPT-5.6 Luna | OpenAI | 1.05M | $0.2 → $1.2 | |
| Grok 4.5 | xAI | 500K | $2.00 → $6.00 | |
| GPT-Live-1 mini | OpenAI | — | — | |
| GPT-Live-1 | OpenAI | — | — | |
| Muse Image | Meta | — | — | |
| GPT-Realtime-2.1 mini | OpenAI | 125K | — | |
| GPT-Realtime-2.1 | OpenAI | 125K | — | |
| Leanstral 1.5 | Mistral AI | — | — | |
| Claude Sonnet 5 | Anthropic | 1M | $2.00 → $10.00 | |
| Gemini Omni Flash Preview | 1M | $1.5 → — | ||
| Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite) | 64K | $0.25 → — | ||
| Qwen-AgentWorld-35B-A3B | Alibaba (Qwen) | 256K | — | |
| Mistral OCR 4 | Mistral AI | — | — | |
| GLM-5.2 | Z.AI | 1M | $1.4 → $4.4 | |
| Claude Mythos 5 | Anthropic | 1M | $10.00 → $50.00 | |
| Claude Fable 5 | Anthropic | 1M | $10.00 → $50.00 | |
| Qwen3.7-Plus | Alibaba (Qwen) | 1M | $0.4 → $1.6 | |
| MiniMax M3 | MiniMax | 1M | $0.3 → $1.2 | |
| Grok Build 0.1 | xAI | 250K | $1.00 → $2.00 | |
| Claude Opus 4.8 | Anthropic | 1M | $5.00 → $25.00 | |
| Gemini 3.5 Flash | 1M | $1.5 → $9.00 | ||
| Mistral Medium 3.5 | Mistral AI | 250K | $1.5 → $7.5 | |
| DeepSeek-V4-Pro | DeepSeek | 1M | $1.32 → $3.96 | |
| DeepSeek-V4-Flash | DeepSeek | 1M | $0.44 → $1.32 | |
| GPT-5.5 | OpenAI | 1.05M | $5.00 → $30.00 | |
| Qwen3.6-27B (dense) | Alibaba (Qwen) | 256K | — | |
| ChatGPT Images 2.0 (GPT Image 2) | OpenAI | — | — | |
| Qwen3.6-35B-A3B (MoE) | Alibaba (Qwen) | 256K | — | |
| Claude Opus 4.7 | Anthropic | 1M | $5.00 → $25.00 | |
| Gemini 3.1 Flash TTS Preview | 8K | — | ||
| Qwen3.6-Plus | Alibaba (Qwen) | 1M | $0.5 → $3.00 | |
| Gemma 4 31B IT | 250K | — | ||
| Gemma 4 26B A4B IT | 250K | — | ||
| Mistral Small 4 | Mistral AI | 250K | $0.15 → $0.6 | |
| Grok 4.20 | xAI | 1M | $1.25 → $2.5 | |
| GPT-5.4 | OpenAI | 1.05M | $2.5 → $15.00 | |
| Gemini 3.1 Pro | 1M | $2.00 → $12.00 | ||
| Claude Sonnet 4.6 | Anthropic | 1M | $3.00 → $15.00 | |
| Qwen3.5-397B-A17B | Alibaba (Qwen team) | 256K | — | |
| Claude Opus 4.6 | Anthropic | 1M | $5.00 → $25.00 |
September 2026
- Grok Voice Transcribe 2.0 — September 18, 2026. xAI added
grok-voice-transcribe-2.0to its speech-to-text API beside the existing model rather than in place of it:grok-voice-transcribe-1.0is still the default, and xAI says 2.0 will take that position soon. Billing is by audio, not by token, and did not move — $0.10 an hour over REST, $0.20 an hour streaming, the same as 1.0. xAI describes dozens of languages, automatic detection, and switching language mid-recording, and reports word error rate falling from 20.6% to 6.8% on its own short-phrase multilingual sets, with telephony audio at 8 kHz its strongest result. Those figures are xAI's, as is the claim of being twice as accurate as 1.0; benchr has tested none of it. Pricing source: xAI. - Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking — September 15, 2026. Google made both voice models generally available through the Live API.
gemini-3.8-liveis the one Google recommends for most low-latency voice agents;gemini-3.8-live-extended-thinkingis for interactions that need heavier background reasoning, and it behaves differently in three documented ways — function calling is asynchronous only and a blocking call errors, the thinking level must be low, medium or high with minimal unsupported, and proactive audio is permanently on. Both take text, images, audio and video and return text and audio, both list 131,072 input and 65,536 output tokens, and both are priced together: $0.75 input and $4.50 output per 1M text tokens, $3.00 and $12.00 per 1M audio tokens, which Google also quotes as $0.005 and $0.018 a minute. Caching, code execution, structured outputs and the Batch API are documented as unsupported. Google published no benchmark table. Pricing source: Google. - Fugu Max and Fugu Ultra v2 — September 11, 2026. Sakana AI is new to benchr's record, and what it ships is not a model in the usual sense. Fugu is an orchestrator: one request to one OpenAI-compatible endpoint, and it routes the work across a pool of other models and assembles the answer. Fugu Max is the cost tier at $2 in and $6 out for every million tokens, with cached input at $0.25 and one rate at any context length. Fugu Ultra v2 is the capability tier at $5 in and $30 out, cached input $0.50, rising to $10, $45 and $1.00 above 272K tokens of context. Sakana's built-in web search and fetch tools bill at $0.007 a call on top of tokens, the models are not offered in the EU or the EEA, and Sakana gives Fugu Ultra v2's training cutoff as August 28, 2026. It publishes no context window and no maximum output for either, so benchr records neither rather than reading a limit out of the pricing threshold. The benchmark figures in its post — 48.3 on Chartography, 74.3 on DeepSWE — are Sakana's own. Pricing source: Sakana AI.
- GPT-6 Astra — September 3, 2026. OpenAI shipped
gpt-6-astraand calls it its most capable model, aimed at reasoning, coding, computer use, research and document creation. The model directory lists a 1.05M context window and an April 30, 2026 knowledge cutoff, and publishes no maximum-output figure, so benchr records none. Pricing is $10.00 input / $1.00 cached input / $50.00 output per 1M tokens, with Batch at half of each — 2.5x GPT-5.6 Sol on both headline rates. The changelog records interface constraints that are migration blockers rather than footnotes: nononereasoning-effort level, no custom temperature,top_porlogprobs, and tool calling only through the Responses API. OpenAI published no benchmark table with the release. Pricing source: OpenAI. - Gemini 3.8 Flash — September 2, 2026. Google made
gemini-3.8-flashgenerally available with a 1M context window and 64K maximum output. Introductory pricing is $0.75 input / $3.75 output per 1M tokens through December 31, 2026, with cached input at $0.075; every tier doubles on January 1, 2027, batch, flex, priority and cached input included. A free API tier is offered. Gemini 3.8 Flash Cyber is a specialised variant of the same foundational model available only through Google’s Fairwind Program, is not separately priced in the API documentation, so benchr carries no record for it. Google states 3.7 Flash remains fully supported. Pricing source: Google · benchr review. - Muse Spark 1.3 — September 2, 2026. Meta released
muse-spark-1.3through the Meta Model API and Muse Code, tuned for agentic work — multi-step tool use, browser use, long-horizon tasks — with coding improved over 1.2. The price did not move: $1.25 in and $4.25 out for every million tokens, cached input $0.15, unchanged across 1.1, 1.2 and 1.3. A separate contributor identifier,muse-spark-1.3-contributor, drops that to $0.10 and $0.20 in exchange for permission to train on your prompts and completions; the standard tier is not used to train Meta models. The model table documents text, image, video, audio and PDF input with text output, a 1,048,576-token window, limited audio support in 1.3, and a “max” reasoning effort this version alone accepts. Meta publishes no maximum output and shows its comparison against GPT‑5.6 Sol and Claude Opus 5 as a figure rather than a table, so benchr records no score from it. This release also closed a gap in benchr's own coverage: Muse Spark 1.2, from August 5, was recorded in the same pass. Pricing source: Meta. - Claude Fable 5.1 and Claude Mythos 5.1 — September 1, 2026. Anthropic introduced
claude-fable-5-1and the access-restrictedclaude-mythos-5-1. Both list 1M context, 128K maximum output, adaptive thinking, $10/$50 standard input/output per 1M, $0.25 cache reads, and Batch $5/$25. Fable 5.1 is generally available through supported Anthropic and partner-cloud paths; Mythos 5.1 is limited to vetted Project Glasswing access. The migration contract changes: named or required tool choice, manual thinking controls, and assistant prefills return errors. Source: Anthropic · pricing.
August 2026
- Qwen3.8-Flash — August 27, 2026. Alibaba released a multimodal mixture-of-experts model: a 125B main model with a further 51B of N-gram embeddings, activating 6B parameters per token, with a native 262K context the announcement says extends to 1M. Model Studio lists it at $0.15 input and $0.47 output per 1M tokens in the Singapore region. Alibaba names SWE-bench Pro, CoWorkBench, Toolathlon Verified, MathVision, AndroidWorld and ERQA in the release but publishes no scores for them, and names no license, so benchr records neither. Pricing source: Alibaba Cloud.
- GLM-5.3-Flash — August 26, 2026. Z.ai released a natively multimodal mixture-of-experts model with 320B total and 18B active parameters, a 1M context window and 128K maximum output, with weights published under MIT. A launch promotion halves every token rate through September 9, 2026: $0.075 input / $0.25 output / $0.015 cached input per 1M tokens, against a list rate of $0.15 / $0.50 / $0.03. The published Terminal-Bench 2.1, DeepSWE and Humanity’s Last Exam figures are Z.ai’s own numbers, measured at a 300K context rather than the full API window, and are not benchr tests. Pricing source: Z.ai.
- DeepSeek-V4-Flash-Vision-Exp — August 21, 2026. DeepSeek published an experimental vision checkpoint called with
deepseek-v4-flash-vision-exp, listing a 1M context window, 384K maximum output and 2,500 concurrent requests. The pricing table places it alongside DeepSeek-V4-Flash at the same rates — $0.44 input / $1.32 output per 1M tokens, halved off-peak, with cache hits at $0.014 — and images are tokenized by dimension and billed as input. DeepSeek labels the checkpoint experimental, so it can change or disappear without a deprecation notice, and no license or open-weights release was found for it on an official page. The published Terminal-Bench 2.1, DeepSWE and Chartography figures are DeepSeek’s own numbers for this checkpoint, not benchr tests. Pricing source: DeepSeek. - GPT-5.6 Sol price cut — August 21, 2026. OpenAI dropped Sol from $5/$30 to $4 input and $20 output per 1M tokens, a 20% cut on input and 33% on output, with cached input falling to $0.40. The changelog calls the rate promotional and available at least through November 21, 2026. Source: OpenAI. benchr pricing.
- OpenAI Model Spec update — On August 18, OpenAI updated the Model Spec. This is a behavior-policy change, not a new production model, so BenchR records it here without inventing an API SKU. Official source.
- GLM-5.3 — Z.AI released the hosted
glm-5.3endpoint on August 18 with 1M context, 128K maximum output, always-on reasoning, and $1.40/$4.40 list pricing. The provider announced that weights would follow after safety hardening; they are not recorded as available yet. BenchR analysis · Official source. - Grok 4.6 in GitHub Copilot — xAI announced Grok 4.6 availability in GitHub Copilot on August 14, two days after the API launch. Official source.
- Gemini 3.7 Flash — Generally available August 13, 2026. Google calls
gemini-3.7-flashits most intelligent workhorse model for coding and agents, with a 1,048,576-token input limit and 65,536-token output limit. Introductory pricing is $0.75 input and $3.75 output per 1M tokens through December 31, 2026, then $1.50/$7.50 — the same rate Gemini 3.6 Flash was cut to. Google published no benchmark table. Source: Google. benchr analysis · model review. - Grok 4.6 — xAI launched
grok-4.6on August 12 with a 500K context window, text and image input, tool use, and base pricing of $2 input, $6 output, and $0.50 cached input per million tokens. BenchR analysis · Official source. - Qwen3.8-Max — August 3, 2026. Alibaba announced its largest flagship to date: a sparse mixture-of-experts model with 2.4T total parameters activating 95B per token, multimodal across text and vision, with a context window the announcement describes as up to 1M tokens. Model Studio lists $2.00 input and $6.00 output per 1M tokens in the Singapore region. Alibaba published arena placements rather than numeric scores — fifth in Text Arena, second in Vision Arena, fourth in Frontend Code Arena — and a placement is not a score, so benchr records no benchmark figures. Weights were said to follow the next week; benchr has verified no license on an official page. Pricing source: Alibaba Cloud.
July 2026
- Kimi K3 — Moonshot AI released Kimi K3 on July 16 and later published the weights. The official record lists 2.8T total and 104B active parameters, 1,048,576 context tokens, and hosted pricing of $3 cache-miss input, $0.30 cache-hit input, and $15 output per million. BenchR analysis · Official source.
- Claude Opus 5 — Anthropic released Claude Opus 5 on July 24, 2026. The article covers its verified API record and migration implications. benchr analysis · Official source.
- GPT-Realtime-2.1 — OpenAI announced GPT-Realtime-2.1 on July 6, 2026 for low-latency voice and multimodal experiences. benchr analysis · Official source.
- GPT-Realtime-2.1 mini — OpenAI announced GPT-Realtime-2.1 mini on July 6, 2026 as its lower-cost realtime companion. benchr analysis · Official source.
- GPT-Live-1 — OpenAI introduced GPT-Live-1 on July 8, 2026 for ChatGPT Voice and said API availability was planned. benchr analysis · Official source.
- GPT-Live-1 mini — OpenAI introduced GPT-Live-1 mini on July 8, 2026 as part of the ChatGPT Voice rollout. benchr analysis · Official source.
- Gemini 3.5 Flash-Lite — Google released Gemini 3.5 Flash-Lite as generally available on July 21, 2026. benchr analysis · Official source.
- Leanstral 1.5 — Mistral released Leanstral 1.5 on July 2, 2026 as an Apache-2.0 formal-verification specialist. benchr analysis · Official source.
- Qwen-Audio 3.0 TTS Plus — Alibaba Cloud released Qwen-Audio 3.0 TTS Plus on July 14, 2026. benchr analysis · Official source.
- Qwen-Audio 3.0 TTS Flash — Alibaba Cloud released Qwen-Audio 3.0 TTS Flash on July 14, 2026. benchr analysis · Official source.
- Muse Image — Meta launched Muse Image on July 7, 2026 and made it available in Meta AI. benchr analysis · Official source.
- Muse Video Preview — Meta previewed Muse Video on July 7, 2026 alongside Muse Image. benchr analysis · Official source.
- Gemini 3.6 Flash — Released July 21, 2026 as a stable Gemini API model. Google lists
gemini-3.6-flashat $1.50 input, $7.50 output, and $0.15 cached input per 1M tokens, with a 1,048,576-token input window and 65,536-token output limit. Source: Google. benchr analysis. - Meta Muse Spark 1.1 — Released July 9, 2026 in public preview through Meta Model API. Meta positions Spark as a 1M-token multimodal model for long-context creative workflows; no public API price was posted in the announcement, so benchr tracks it in model-figures.json but keeps it out of the ranked pricing tool for now. Source: Meta.
- Grok 4.5 — Released to the xAI API July 8, 2026. xAI lists
grok-4.5with a 500K context window at $2 input, $6 output, and $0.50 cached input per 1M tokens. xAI does not publish official benchmark tables or a max-output limit for this model yet. benchr review · Source: xAI · benchr pricing. - GPT-5.6 — generally available — July 9, 2026. OpenAI's API changelog now lists GPT-5.6 Sol, Terra, and Luna as released to the API. Official prices are unchanged from the June preview: Sol $5/$30, Terra $2.50/$15, Luna $1/$6 per 1M; OpenAI's models page lists all three with 1.05M context and 128K max output. Source: OpenAI. benchr writeup · pricing.
- Claude Sonnet 5 — Released July 1, 2026. Anthropic's second Mythos-class-architecture model after Claude Fable 5, priced at $2/$10 per million tokens — a launch rate Anthropic confirmed as standard on August 31 after cancelling the September rise to $3/$15 — with a 1M context window and a 128K max output — double Sonnet 4.6's 64K. Anthropic reports SWE-bench Verified 89.4% and GPQA Diamond 92.0%. Safety-classified requests return an explicit refusal; another-model retry requires configured application or product logic. Source: Anthropic. benchr coverage · pricing.
- Claude Fable 5 restored to all customers — July 1, 2026. The U.S. Commerce Department export-control review that suspended Fable 5 on June 12 concluded, and Anthropic restored access for every customer, with AWS restoring Bedrock access the same day; Claude Mythos 5 remains limited to Project Glasswing. Landed the same day as the Claude Sonnet 5 launch above. Source: Anthropic. benchr coverage.
June 2026
- Gemini Omni Flash Preview — Released June 30, 2026. Google's new preview model accepts text, image, video, and audio input and can produce text plus short video clips. The Gemini docs list a 1,048,576-token context window and pricing of $1.50 input, $9 text output, and $17.50 video output per 1M tokens. Source: Google.
- Gemini 3.1 Flash Lite Image — Released June 30, 2026. Google lists this Nano Banana 2 Lite image model as generally available with a 65,536-token context window, 4,096-token max output, $0.25 input, $1.50 text output, and $30 image output per 1M tokens. Source: Google.
- GLM-5.2 — Released June 16, 2026 by Z.AI. The API model
glm-5.2lists a 1M-token context window, 128K max output, and pricing at $1.40 input, $4.40 output, and $0.26 cache-hit input per 1M tokens. benchr review · Source: Z.AI · benchr pricing. - MiniMax M3 — Released June 1, 2026. MiniMax lists
MiniMax-M3as a 1M-context text model with standard pricing at $0.30 input, $1.20 output, and $0.06 cache-hit input per 1M tokens; long-context and priority tiers cost more. Source: MiniMax. benchr pricing. - GPT-5.6 — Sol, Terra, and Luna — Previewed June 26, 2026, then generally available July 9. The preview introduced Sol, Terra, Luna, and the max/ultra reasoning modes under government-gated partner access; the July 9 API changelog moved the family into general availability. Source: OpenAI.
- Mistral OCR 4 — Released June 23, 2026. State-of-the-art document OCR: paragraph-level bounding boxes, typed-block labels, 170 languages. Priced per page ($4 / 1,000 pages standard, $2 batch, $5 Document AI), not per token. Source: Mistral.
- Qwen-AgentWorld (35B-A3B and 397B-A17B) — Released June 24, 2026. Open-weight (Apache-2.0) “language world models” for agent-environment simulation across seven domains; 256K context, download-only. Source: Qwen.
- Kimi K2.7-Code — Moonshot's coding-specialist successor to K2.6, built on the same model. $0.95/1M input (cache miss), $0.19 cache hit, $4/1M output, 256K context, Modified MIT; a Highspeed tier doubles the rates ($1.90/$8) for ~180 tok/s. Moonshot publishes no official release date. Source: Moonshot. benchr pricing.
- Claude Fable 5 and Claude Mythos 5 — Released June 9, 2026. Fable 5 lists at $10/$50 per million tokens with 1M context and 128K max output; safety-classified requests return an explicit refusal, and a retry on another model must be configured. Mythos 5 stays restricted to Project Glasswing. Source: Anthropic. benchr coverage. Timeline: both models were suspended June 12, controls were lifted June 30, and access returned July 1. Fable 5 is generally available worldwide; Mythos 5 remains limited to approved Glasswing organizations.
May 2026
- Grok Build 0.1 — Reached the xAI API in public beta around May 29, 2026. Agentic coding model, 256K context, $1.00/$2.00 per 1M. Source: xAI.
- Qwen3.7-Max — Announced May 21, 2026 and now available through Alibaba Cloud Model Studio. The announcement still described API access as “coming soon,” so benchr does not assign an unsupported first-availability date. Singapore/international list pricing is $2.50/$7.50 per 1M tokens with a 1M input tier. Announcement; current availability; pricing.
- Claude Opus 4.8 — Released May 28, 2026. Anthropic's newest flagship. Same $5/$25 pricing as 4.7, 69.2% on SWE-Bench Pro (up from 64.3%), with a notable honesty and alignment gain. benchr review.
- Gemini 3.5 Flash — Released May 19, 2026. Beats Gemini 3.1 Pro on most coding and agentic benchmarks. $1.50 input / $9 output per million tokens. 1M context. Source: Google. benchr review.
April 2026
- Gemini 3.1 Flash TTS Preview — Google launched Gemini 3.1 Flash TTS Preview on April 15, 2026. benchr analysis · Official source.
- Gemma 4 26B A4B IT — Google released gemma-4-26b-a4b-it on April 2, 2026 as part of the Gemma 4 launch. benchr analysis · Official source.
- Gemma 4 31B IT — Google released gemma-4-31b-it on April 2, 2026 as part of the Gemma 4 launch. benchr analysis · Official source.
- Mistral Medium 3.5 — Released April 28, 2026. 128B dense, 256K context, vision built-in. Replaces Medium 3.1 + Magistral + Devstral 2 in one model. $1.50 / $7.50 per million tokens. Modified MIT license. Source: Mistral. benchr review.
- DeepSeek V4-Pro and V4-Flash — Released April 24, 2026. V4-Pro: 1.6T MoE, 49B active, 80.6% SWE-bench Verified. V4-Flash: 284B MoE, 13B active, $0.14 input at launch, repriced to $0.44 peak / $0.22 off-peak on August 16, 2026. Source: DeepSeek. benchr review.
- GPT-5.5 — Released April 23, 2026. OpenAI's new flagship. $5 input / $30 output. 1M context. Strong on Terminal-Bench and FrontierMath. Source: OpenAI. benchr review.
- Claude Opus 4.7 — Released April 16, 2026. Same $5/$25 pricing as Opus 4.6, 87.6% SWE-bench Verified. New tokenizer produces up to 35% more tokens per request. Source: Anthropic. benchr review.
March 2026
- Mistral Small 4 — Released March 16, 2026. Open-weight (Apache-2.0) hybrid MoE, 119B total / 6B active, 256K context, text + image input. $0.15/$0.60 per 1M. Source: Mistral.
- GPT-5.4 family — Released March 5, 2026 (Thinking + Pro). GPT-5.4 mini and nano followed on March 17. $2.50/$15 for the main model, context up to 1M tokens, built-in computer use. Launched alongside ChatGPT for Excel. Source: OpenAI.
- GPT-5.3 Instant — Released March 3, 2026 as the ChatGPT default model. Replaced by GPT-5.5 Instant on May 5; paid users keep it for roughly three months from that date. Source: OpenAI release notes.
- Gemini 3 Pro deprecated — March 9, 2026. Replaced by Gemini 3.1 Pro Preview as the Pro tier. Source: Google changelog.
February 2026
- Gemini 3.1 Pro Preview — Released February 19, 2026. $2 input / $12 output per million tokens. Still the Pro tier; Gemini 3.5 Pro remains unreleased as of this writing. benchr review.
- Qwen 3.5 — Released February 16, 2026. Flagship 397B MoE, 17B active; the family spans nine sizes, most under Apache 2.0. Superseded by Qwen3.6 in April. Source: Qwen GitHub. benchr review.
What to watch
Claude Fable 5's return — resolved. The export-control review Commerce opened in June concluded on July 1, 2026, and Anthropic restored Fable 5 to all customers, with AWS restoring Bedrock access the same day. Claude Mythos 5 remains limited to Project Glasswing.
Next up. OpenAI has disclosed an upcoming model named Astra in a safety-planning announcement, but has not released a production model card, API identifier, or price. It remains a watch item rather than a BenchR model record. Z.AI's GLM-5.3 weights also remain behind the announced safety-hardening period, and Moonshot's Kimi K2.5/Moonshot V1 platform sunset is scheduled for August 31.
The open-weight tier (DeepSeek V4, Qwen 3.5/3.6, Mistral Medium 3.5) has closed the gap with closed labs to single-digit benchmark points on most evaluations. Coding and reasoning leaders are now distributed across both camps.
Changelog
- — Added Fable 5.1 and Mythos 5.1 from Anthropic's official sources, with bilingual migration and pricing coverage.
- — Added the August official-source sweep, the July Kimi K3 record, dated lifecycle changes, and links to the four bilingual analyses.