Author's Description
Morph's high-accuracy apply model for complex code edits. 2000+ tokens/sec with 98% accuracy for precise code transformations.
Key Specifications
Supported Parameters
This model supports the following parameters:
Performance Summary
Morph V3 Large, created by morph on July 7, 2025, is designed as a high-accuracy model for complex code edits, boasting a throughput of 2000+ tokens/sec with 98% accuracy for precise code transformations. In terms of speed, the model consistently ranks among the fastest, achieving an Infinityth percentile across one benchmark, indicating exceptional processing speed. Price competitiveness cannot be assessed due to the absence of cost data, suggesting potential free tier usage. A significant concern is the model's reliability, which is notably low, ranking at the 0th percentile across one benchmark. This indicates frequent technical failures, including network timeouts, API errors, and system errors, severely impacting its usability. Performance on the "Coding (Baseline)" benchmark was 0.0% accuracy over a duration of 133498ms, which is a critical weakness given its intended application for code edits. While its speed is a clear strength, the severe reliability issues and complete lack of accuracy in the tested coding benchmark represent major weaknesses that undermine its utility despite its promising description.
Model Pricing
Current Pricing
Feature | Price (per 1M tokens) |
---|---|
Prompt | $0.9 |
Completion | $1.9 |
Price History
Available Endpoints
Provider | Endpoint Name | Context Length | Pricing (Input) | Pricing (Output) |
---|---|---|---|---|
Morph
|
Morph | morph/morph-v3-large | 81K | $0.9 / 1M tokens | $1.9 / 1M tokens |
Benchmark Results
Benchmark | Category | Reasoning | Free | Executions | Accuracy | Cost | Duration |
---|
Other Models by morph
|
Released | Params | Context |
|
Speed | Ability | Cost |
---|---|---|---|---|---|---|---|
Morph: Morph V3 Fast | Jul 07, 2025 | — | 81K |
Text input
Text output
|
— | ★ | $$$$ |
Morph: Fast Apply Unavailable | Jun 26, 2025 | — | 32K |
Text input
Text output
|
— | ★ | $$$$$ |