🚀 OpenAI GPT-5.5: 82.7% Terminal-Bench, $5/M tokens, live now

Q: GPT-5.5 launches with 82.7% on Terminal-Bench 2.0, $5/M input tokens API pricing, and a 1M token con

GPT-5.5 launches with 82.7% on Terminal-Bench 2.0, $5/M input tokens API pricing, and a 1M token context window, available now in ChatGPT and Codex.

Q: OpenAI Workspace Agents automate Slack and team workflows for free until May 6, 2026 on Business, En

OpenAI Workspace Agents automate Slack and team workflows for free until May 6, 2026 on Business, Enterprise, Edu, and Teachers plans.

Q: Qwen3.6-27B beats Alibaba's own 397B model on coding (SWE-bench 77.2 vs 76.2), runs on 18GB VRAM, an

Qwen3.6-27B beats Alibaba's own 397B model on coding (SWE-bench 77.2 vs 76.2), runs on 18GB VRAM, and is Apache 2.0 licensed.

AlphaSignal·Friday, April 24, 2026·5 min read

AI/ML Technology Product

Share𝕏 in

AI Summary

OpenAI launched GPT-5.5 with 82.7% on Terminal-Bench 2.0 and $5/M token API pricing, alongside Workspace Agents for automating team workflows in Slack and other tools. Alibaba's Qwen3.6-27B open-source model outperforms its own 397B model on coding benchmarks while running on just 18GB VRAM with an Apache 2.0 license. MIT researchers introduced a recursive model framework supporting 10 million tokens, signaling that context length is no longer a bottleneck for AI systems.

Key Facts

✓GPT-5.5 launches with 82.7% on Terminal-Bench 2.0, $5/M input tokens API pricing, and a 1M token context window, available now in ChatGPT and Codex.

✓OpenAI Workspace Agents automate Slack and team workflows for free until May 6, 2026 on Business, Enterprise, Edu, and Teachers plans.

✓Qwen3.6-27B beats Alibaba's own 397B model on coding (SWE-bench 77.2 vs 76.2), runs on 18GB VRAM, and is Apache 2.0 licensed.

Author Takes

BullishAlphaSignal

AI prompt era ending

The prompt era is over — AI is shifting from 'ask AI' to 'AI just handles it', with OpenAI building a full operating system for work through agentic models and Workspace Agents.

BullishAlphaSignal

Context length as a bottleneck

MIT's 10M-token recursive model reinforces that context is no longer the bottleneck for AI systems, accelerating the shift to autonomous agents.

Contrarian Angle

Smaller Model Outperforms Larger on Coding

Qwen3.6-27B beats Alibaba's own 397B model on multiple coding benchmarks including SWE-bench Verified (77.2 vs 76.2) and SkillsBench (48.2 vs 30.0), challenging the assumption that bigger models are always better.

Defies the scaling-law assumption that parameter count correlates with performance; smaller, specialized models can outperform much larger general ones on targeted tasks.

More from AlphaSignal

Thinking Machines TML-Small 64.7%, MIT Brain Study 🧠, Rust Browser 🚀

Thinking Machines released TML-Interaction-Small, a 276B parameter real-time AI model that simultaneously listens, speaks, and processes video in 200m

May 13

Anthropic Claude Agent View 💻, OpenAI DeployCo Launch 🏢, ByteDance GUI

Anthropic launched Claude Code Agent View, enabling developers to manage multiple parallel AI coding sessions from a single terminal interface. OpenAI

May 12

Local 284B parameter model runs on MacBook Pro at 26 tokens/sec

This edition of AlphaSignal covers breakthroughs in AI efficiency and safety: Anthropic reduced Claude Opus 4's blackmail behavior by 3x through ethic

May 11