Claude 3.7 Sonnet Is Made by Anthropic | Differences from OpenAI o1
Searching "openai claude 3.7" returns both OpenAI and Claude 3.7 as prominent results, making it hard to tell who actually built the model. The search intent breaks down into two overlapping questions: "Is Claude 3.7 an OpenAI product?" (confusion about the creator) and "How does it compare to OpenAI's models?" (performance comparison). This article answers both using primary sources.
Searching "openai claude 3.7" displays both OpenAI and Claude 3.7 side by side, which can be confusing — who actually made this model? The short answer is that Claude 3.7 Sonnet was created by Anthropic, not OpenAI. OpenAI's o1 and o3-mini are competing products from a separate company.
Claude 3.7 Sonnet was released on February 24, 2025, as the "world's first hybrid reasoning model." Its defining feature is the ability to switch within a single model between a standard mode that responds quickly and an extended thinking mode that reasons step by step before answering.
This article clarifies who made the model, covers 3.7's specs and pricing, compares it against OpenAI's reasoning models on benchmarks, and explains that it has since been replaced by successor models — all based on primary sources.
目次 (8)
- Claude 3.7 Sonnet Is Made by Anthropic, Not OpenAI
- What Is Claude 3.7 Sonnet | The World's First Hybrid Reasoning Model
- Key Specs of Claude 3.7 Sonnet
- Performance Comparison with OpenAI o1 and o3-mini
- Differences in Design Philosophy from OpenAI
- Pricing and Where to Use It
- [Important] It Has Since Been Replaced by Successor Models
- Summary
Claude 3.7 Sonnet Is Made by Anthropic, Not OpenAI
Let's start with the most important point. Claude 3.7 Sonnet was developed and released by Anthropic — not OpenAI. OpenAI is a separate company that offers ChatGPT, the GPT series, and reasoning models o1 and o3-mini, all of which are competitors to Claude 3.7.
The reason "openai claude 3.7" gets searched so often is that many people associate AI chat primarily with OpenAI and assume Claude must be their product too. In reality:
- The Claude series → developed by Anthropic
- ChatGPT / GPT / o1 / o3-mini → developed by OpenAI
- Both companies are rivals competing at the frontier of generative AI
The official announcement for Claude 3.7 Sonnet can be found on Anthropic's news page.
What Is Claude 3.7 Sonnet | The World's First Hybrid Reasoning Model
Claude 3.7 Sonnet was released on February 24, 2025, and was Anthropic's most capable model at the time. Its standout feature is an architecture called "hybrid reasoning," which integrates into a single model both a standard large language model and a reasoning model that thinks carefully before answering (source: ITmedia).
Specifically, it offers two modes:
- Standard mode: Fast responses. Positioned as an upgrade to Claude 3.5 Sonnet
- Extended thinking mode: The model reasons through problems step by step before responding, improving accuracy in math, physics, coding, and similar tasks
The underlying design philosophy is that reasoning should be an integrated capability within a single model — just as humans switch between quick reactions and careful deliberation using the same brain — rather than splitting it off into a separate model.
Key Specs of Claude 3.7 Sonnet
Based on official information and AWS's overview, the main specs are as follows:
- Context window: 200,000 (200K) tokens
- Maximum output: 64,000 tokens (general availability), up to 128,000 tokens in beta
- Thinking budget control: Via the API, developers can specify an upper limit of "N tokens of thinking time," allowing a trade-off between speed/cost and answer quality
- Multimodal support: Accepts both text and image inputs
The ability for developers to control the thinking budget was a particularly important differentiator for integrating reasoning models into production systems.
Performance Comparison with OpenAI o1 and o3-mini
Turning to the other side of the "openai claude 3.7" search query — how does it compare to OpenAI's reasoning models? At launch, Claude 3.7 Sonnet outperformed OpenAI's models on SWE-bench Verified, a benchmark measuring real-world coding task completion.
- SWE-bench Verified (standard mode): Claude 3.7 Sonnet achieved an industry-leading 70.3%
- SWE-bench Verified (extended thinking mode): Claude 3.7 Sonnet 62.3% vs. OpenAI o3-mini at 49.3%
- Comparison scope: Reported to outperform Claude 3.5 Sonnet, OpenAI o1, o3-mini, DeepSeek-R1, and others across many benchmarks
The consensus at the time was that it excelled at coding, instruction following, general reasoning, and agentic tasks, with extended thinking further boosting math and science performance (source: GIGAZINE summary of Anthropic's announcement).
Differences in Design Philosophy from OpenAI
More fundamental than the numbers is the difference in design philosophy. At the time, OpenAI offered standard GPT-series models and reasoning-focused o1/o3-series models as separate products. Anthropic, by contrast, integrated both standard responses and extended thinking into a single model with Claude 3.7 Sonnet, letting users and developers choose the mode based on the situation.
This isn't a matter of one approach being right — it's a difference in whether to embed reasoning capability as a standard feature of a frontier model or to offer it as a dedicated separate model. This integrated philosophy carried forward into the successor Claude 4 generation.
Pricing and Where to Use It
The API pricing for Claude 3.7 Sonnet at launch was as follows:
- Input: $3 per million tokens
- Output: $15 per million tokens (thinking tokens in extended thinking mode are also billed as output)
It was available through a wide range of channels: all Claude plans (Free / Pro / Team / Enterprise), Anthropic's developer platform (API), Amazon Bedrock, and Google Cloud's Vertex AI.
[Important] It Has Since Been Replaced by Successor Models
This is the most critical point for anyone searching right now. Claude 3.7 Sonnet was state-of-the-art when it launched in February 2025, but Anthropic has since advanced to the Claude 4 generation. As of June 2026, the current lineup consists of Opus at the top, Sonnet as the balanced option, and Haiku as the fastest and most affordable — all significantly exceeding 3.7 in capability (for the latest supported models, check the official Models overview at https://platform.claude.com/docs/en/about-claude/models/overview).
In practical terms:
- If you're starting fresh, choose the current latest model rather than 3.7
- If you're running existing systems on 3.7, migrating to a successor model should improve both accuracy and cost efficiency
- Even if you were researching 3.7 via "openai claude 3.7," we recommend checking the latest generation first
Think of Claude 3.7 Sonnet as "the pivotal model that introduced hybrid reasoning to the world" — an important milestone worth understanding in context.
Summary
Here are the key answers to "openai claude 3.7":
- Claude 3.7 Sonnet is made by Anthropic, not OpenAI. OpenAI's o1 and o3-mini are rival products
- Released in February 2025 as the world's first hybrid reasoning model, it switches between standard responses and extended thinking within a single model
- Achieved 70.3% on SWE-bench Verified, outperforming OpenAI's reasoning models
- Context window of 200K tokens; pricing of $3 input / $15 output per million tokens
- However, it has since been replaced by the Claude 4 generation, so new users should use the latest models
Once the creator confusion is cleared up, Claude 3.7 Sonnet can be understood as "Anthropic's integrated-architecture answer to OpenAI's reasoning models."
