Developer Tools Digest: Anthropic 300K Output API, OpenAI GPT-4o Retired, $122B Fundraise, 2026-04-06
ai

Developer Tools Digest: Anthropic 300K Output API, OpenAI GPT-4o Retired, $122B Fundraise, 2026-04-06

5 min read

Anthropic API: 300K Output Tokens on Message Batches, 1M Context Window Retirement Notice

Anthropic has raised the max_tokens cap on the Message Batches API to 300K tokens for Claude Opus 4.6 and Sonnet 4.6, effective March 30, 2026. To activate the extended limit, include the output-300k-2026-03-24 beta header in your request. This is particularly useful for long-form content generation, large structured data extraction, and bulk code generation tasks where a single response needs to cover substantial ground without being split across multiple requests.

The extended output is currently limited to the Message Batches API rather than the standard synchronous Messages API, making it best suited for asynchronous pipelines where latency is less critical than output completeness. Developers already using the Batches API for cost savings—which processes requests at 50% of the standard API price—now get additional utility from the same endpoint.

In a separate but related update, Anthropic has announced the retirement of the 1M token context window beta for Claude Sonnet 4.5 and Claude Sonnet 4 on April 30, 2026. After that date, the context-1m-2025-08-07 beta header will have no effect on these models, and requests exceeding the standard 200K-token context window will return an error. Developers relying on 1M context on those models should migrate to Claude Sonnet 4.6 or Claude Opus 4.6, which natively support the 1M context window at standard pricing without any beta header.

Read more — Claude Platform Release Notes


OpenAI GPT-4o Fully Retired; Codex Switches to Pay-as-You-Go Token Billing

OpenAI has completed the retirement of GPT-4o across all plans as of April 3, 2026. Custom GPTs built on GPT-4o in Business, Enterprise, and Edu tiers had access preserved through that date, but the model is now fully sunsetted. Developers and teams still running on GPT-4o need to update their integrations to GPT-5.4 or the appropriate successor model—API calls to retired models return errors with no grace period beyond what was already communicated.

Alongside the model retirement, OpenAI restructured Codex pricing effective April 2, 2026. The previous per-seat licensing model has been replaced with pay-as-you-go token billing. Standard ChatGPT Business seat pricing has been cut from $25 to $20 per month, with Codex-specific usage billed separately based on token consumption. Credits remain the core billing unit, with rates set per million input tokens, cached input tokens, and output tokens. The shift benefits teams with uneven usage patterns—light users pay significantly less than they did under flat seat pricing, while heavy agentic users can now forecast costs against actual consumption data rather than seat counts.

GPT-5.4 mini is now available to Free and Go users via the "Thinking" feature. Separately, OpenAI is expanding ChatGPT integrations with updated Box, Notion, Linear, and Dropbox app connectors that include write capabilities where supported—a step toward more capable cross-platform agentic workflows.

Read more — WinBuzzer


Anthropic Acquires Biotech Startup Coefficient Bio for $400M

Anthropic has acquired Coefficient Bio, a stealth AI biotech startup, for approximately $400 million in stock. Coefficient Bio had fewer than ten employees and was building a platform using AI to assist with drug research planning, clinical regulatory strategy, and identifying new drug opportunities. The team will join Anthropic's healthcare and life sciences group.

The acquisition marks a notable strategic expansion for Anthropic beyond its core developer tooling and general-purpose AI assistant business. Applied AI in drug discovery is one of the higher-value domains for large language models—the combination of long-context reasoning, structured data analysis, and synthesis of regulatory and scientific literature maps well to Claude's strengths. DeepMind's AlphaFold and Google's material science research lab have demonstrated that specialized AI applications in life sciences attract significant investment and produce commercially valuable outputs.

For developers, the near-term impact is indirect—Anthropic's engineering attention and API roadmap are not likely to shift. Longer term, the acquisition may surface new API capabilities oriented toward scientific and regulated-domain workflows, particularly given Anthropic's parallel investments in data residency and compliance controls announced earlier in 2026.

Read more — TechCrunch


OpenAI Raises $122 Billion, Plans IPO as Monthly Revenue Reaches $2 Billion

OpenAI announced on April 2, 2026, that it has raised $122 billion in a new funding round to accelerate its next phase of AI development. The company is reporting $2 billion in monthly revenue and has signaled plans for a public offering that would reshape the AI industry's funding landscape.

The scale of the raise reflects the intensifying race among frontier AI labs. With Anthropic having raised over $10 billion and Google DeepMind operating with effectively unlimited infrastructure through Google Cloud, the capital positions of these three labs dwarf any other players in the market. OpenAI's stated use of funds focuses on continued compute infrastructure build-out, safety research, and expanding distribution through enterprise agreements.

For developers building on OpenAI's APIs, the funding does not directly change product availability or pricing, but it does signal continued investment in model capability improvements and capacity. The planned IPO introduces a longer-term variable—public company governance and quarterly earnings pressure can shift product roadmap priorities in ways that private companies avoid. Teams with multi-year API dependencies on OpenAI's platform may want to track the IPO timeline as a planning variable.

Read more — OpenAI Release Notes via Releasebot


Stanislav Lentsov

Written by

Stanislav Lentsov

Software Architect

You May Also Enjoy