feat: add MiniMax models to the catalog - #180
Conversation
|
All contributors have signed the CLA. Thank you. |
📝 WalkthroughSummary by CodeRabbit
WalkthroughThe catalog now defines MiniMax models, regional endpoints, metadata, and pricing. Documentation describes routing MiniMax through existing Claude and Aider adapters. New tests validate catalog consistency, documentation references, and adapter model resolution. ChangesMiniMax compatibility
Estimated code review effort: 3 (Moderate) | ~20 minutes Suggested reviewers: Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
providers/model_catalog.json (1)
82-99: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winConsider a
notesfield explaining the single-model tier strategy.
latest_developmentandlatest_fastboth resolve toMiniMax-M3(tier"planning"), whileMiniMax-M2.7is tagged"tier": "development"but is never referenced by anylatest_*field. Thecodexprovider block above has the identical asymmetry (latest_development/latest_fast→gpt-5.3-codex, tier"planning") and documents it with a"notes"field. Adding an equivalent note here would prevent future maintainers from assuminglatest_developmentshould point at the model tagged"development".♻️ Suggested addition
"minimax": { "latest_planning": "MiniMax-M3", "latest_development": "MiniMax-M3", "latest_fast": "MiniMax-M3", + "notes": "MiniMax-M3 is used for all tiers; MiniMax-M2.7 (tier: development) is catalog-only and not a latest_* default.", "endpoints": [🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@providers/model_catalog.json` around lines 82 - 99, Add a notes field to the minimax provider configuration documenting that latest_development and latest_fast intentionally resolve to MiniMax-M3 despite its planning tier, while MiniMax-M2.7 is the development-tier model and is not selected by a latest_* field. Match the equivalent documentation pattern used in the codex provider block.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@providers/model_catalog.json`:
- Around line 100-163: Update the max_output values in the models array: set
MiniMax-M3 to 512000 and MiniMax-M2.7 to 128000. Leave all other model metadata
and pricing unchanged.
---
Nitpick comments:
In `@providers/model_catalog.json`:
- Around line 82-99: Add a notes field to the minimax provider configuration
documenting that latest_development and latest_fast intentionally resolve to
MiniMax-M3 despite its planning tier, while MiniMax-M2.7 is the development-tier
model and is not selected by a latest_* field. Match the equivalent
documentation pattern used in the codex provider block.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 213badce-927e-4973-bc0e-3c5913e3c839
📒 Files selected for processing (4)
providers/model_catalog.jsontests/run-all-tests.shtests/test-minimax-model-catalog.shwiki/Providers.md
| "models": [ | ||
| { | ||
| "id": "MiniMax-M3", | ||
| "tier": "planning", | ||
| "context_window": 1000000, | ||
| "max_output": 524288, | ||
| "input_modalities": ["text", "image", "video"], | ||
| "thinking": ["adaptive", "disabled"], | ||
| "pricing_usd_per_million_tokens": { | ||
| "input": 0.3, | ||
| "output": 1.2, | ||
| "cache_read": 0.06, | ||
| "cache_write": null | ||
| }, | ||
| "pricing_tiers_usd_per_million_tokens": [ | ||
| { | ||
| "service_tier": "standard", | ||
| "input_tokens_lte": 512000, | ||
| "input": 0.3, | ||
| "output": 1.2, | ||
| "cache_read": 0.06, | ||
| "cache_write": null | ||
| }, | ||
| { | ||
| "service_tier": "standard", | ||
| "input_tokens_gt": 512000, | ||
| "input": 0.6, | ||
| "output": 2.4, | ||
| "cache_read": 0.12, | ||
| "cache_write": null | ||
| }, | ||
| { | ||
| "service_tier": "priority", | ||
| "input_tokens_lte": 512000, | ||
| "input": 0.45, | ||
| "output": 1.8, | ||
| "cache_read": 0.09, | ||
| "cache_write": null | ||
| }, | ||
| { | ||
| "service_tier": "priority", | ||
| "input_tokens_gt": 512000, | ||
| "input": 0.9, | ||
| "output": 3.6, | ||
| "cache_read": 0.18, | ||
| "cache_write": null | ||
| } | ||
| ] | ||
| }, | ||
| { | ||
| "id": "MiniMax-M2.7", | ||
| "tier": "development", | ||
| "context_window": 204800, | ||
| "max_output": 204800, | ||
| "input_modalities": ["text"], | ||
| "thinking": ["always_on"], | ||
| "pricing_usd_per_million_tokens": { | ||
| "input": 0.3, | ||
| "output": 1.2, | ||
| "cache_read": 0.06, | ||
| "cache_write": 0.375 | ||
| } | ||
| } | ||
| ] |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🌐 Web query:
What is the official maximum output token limit for MiniMax-M3 and MiniMax-M2.7 per MiniMax's API documentation?
💡 Result:
According to MiniMax's official documentation and resources, the maximum output token limits are as follows: For MiniMax-M3, the maximum output token limit is 512,000 tokens [1]. For MiniMax-M2.7, the documentation specifies a context window of 204,800 tokens [2][3] and indicates a maximum output of 128,000 tokens for the M2-series models [4]. Note that in the context of MiniMax's API, the maximum output limit (often configured via the max_tokens or max_completion_tokens parameter) is subject to the total model context window [2][5]. Users should refer to the specific parameters in the API request body—typically max_tokens or max_completion_tokens—when implementing their integrations [2][5][3].
Citations:
- 1: https://www.minimax.io/blog/minimax-m3
- 2: https://platform.minimax.io/docs/api-reference/text-openai-api
- 3: https://platform.minimax.io/docs/api-reference/text-ai-sdk
- 4: https://platform.minimax.io/docs/guides/models-intro
- 5: https://platform.minimax.io/docs/api-reference/text-anthropic-api
🏁 Script executed:
#!/bin/bash
set -euo pipefail
file="providers/model_catalog.json"
# Show the relevant section with line numbers.
sed -n '95,170p' "$file" | cat -nRepository: asklokesh/loki-mode
Length of output: 2760
🌐 Web query:
MiniMax official documentation maximum output tokens MiniMax-M3 512000 524288 MiniMax-M2.7 128000 204800
💡 Result:
Based on official MiniMax documentation and associated technical specifications, the token limits for the specified models are as follows: MiniMax-M3: The context window supports up to 1,000,000 tokens [1][2]. Various technical aggregators and API providers report a maximum output token limit of 512,000 tokens [3][4][5][6]. MiniMax-M2.7: The context window is 204,800 tokens [7][8][9]. Technical documentation and community-maintained specifications typically cite a maximum output token limit of 128,000 tokens [10][11][12]. It is important to note that while official documentation provides clear context window figures [1][2][9], output token limits are frequently managed via the API's max_completion_tokens parameter [7][13], and specific maximums may vary by platform implementation [3][8]. Always verify the specific constraints via the model's metadata endpoint if building production integrations.
Citations:
- 1: https://platform.minimax.io/docs/guides/text-generation
- 2: https://www.minimax.io/models/text/m3
- 3: https://models.dev/models/minimax/MiniMax-M3
- 4: https://models.sulat.com/models/openrouter-minimaxminimax-m3-2e7b4aad
- 5: https://custom.typingmind.com/tools/estimate-llm-usage-costs/openrouter/minimax-m3
- 6: https://f2api.com/models/minimax-m3
- 7: https://platform.minimax.io/docs/api-reference/text-openai-api
- 8: https://togetherai-52386018.mintlify.app/docs/serverless/models
- 9: https://labex.io/llm-price/model/minimax-minimax-m2-7
- 10: https://github.com/langchain-ai/langchain/blob/0e97e85a/libs/partners/huggingface/langchain_huggingface/data/_profiles.py
- 11: https://models.dev/labs/minimax
- 12: https://github.com/zelosleone/minimax-vscode
- 13: https://platform.minimax.io/docs/api-reference/text-chat-openai
Update the MiniMax output limits MiniMax-M3 should list max_output: 512000 instead of 524288, and MiniMax-M2.7 should list max_output: 128000 instead of 204800. The current values overstate the models’ output ceilings and can cause bad request sizing in downstream planning.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@providers/model_catalog.json` around lines 100 - 163, Update the max_output
values in the models array: set MiniMax-M3 to 512000 and MiniMax-M2.7 to 128000.
Leave all other model metadata and pricing unchanged.
|
I have read the CLA Document and I hereby sign the CLA |
Align MiniMax-M3 pricing_usd_per_million_tokens with the supported catalog configuration (input 0.6, output 2.4, cache_read 0.12) and update the corresponding test assertion.
|
Note GitHub couldn't provide a complete incremental comparison for this pull request, so CodeRabbit is performing a full review instead. This review may take a little longer. |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@tests/test-minimax-model-catalog.sh`:
- Around line 17-106: Update the Python invocation wrapping the catalog
validation script to include the -E interpreter option before executing the
heredoc, ensuring environment-based optimization cannot remove the assert checks
in the validation code. Keep the existing catalog parsing and assertions
unchanged.
🪄 Autofix
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 4161c3e9-83df-4ad2-9097-ac45ca736521
📒 Files selected for processing (4)
providers/model_catalog.jsontests/run-all-tests.shtests/test-minimax-model-catalog.shwiki/Providers.md
🚧 Files skipped from review as they are similar to previous changes (2)
- tests/run-all-tests.sh
- wiki/Providers.md
| if python3 - "$CATALOG" <<'PY' | ||
| import json | ||
| import sys | ||
|
|
||
| with open(sys.argv[1], encoding="utf-8") as handle: | ||
| catalog = json.load(handle) | ||
|
|
||
| provider = catalog["providers"]["minimax"] | ||
| assert provider["latest_planning"] == "MiniMax-M3" | ||
| assert provider["latest_development"] == "MiniMax-M3" | ||
| assert provider["latest_fast"] == "MiniMax-M3" | ||
|
|
||
| endpoints = {item["region"]: item for item in provider["endpoints"]} | ||
| assert endpoints == { | ||
| "global_en": { | ||
| "region": "global_en", | ||
| "openai_base_url": "https://api.minimax.io/v1", | ||
| "anthropic_base_url": "https://api.minimax.io/anthropic", | ||
| "docs_root": "https://platform.minimax.io/docs", | ||
| }, | ||
| "cn_zh": { | ||
| "region": "cn_zh", | ||
| "openai_base_url": "https://api.minimaxi.com/v1", | ||
| "anthropic_base_url": "https://api.minimaxi.com/anthropic", | ||
| "docs_root": "https://platform.minimaxi.com/docs", | ||
| }, | ||
| } | ||
| assert all(item["anthropic_base_url"].endswith("/anthropic") for item in endpoints.values()) | ||
|
|
||
| models = {item["id"]: item for item in provider["models"]} | ||
| assert list(models)[:2] == ["MiniMax-M3", "MiniMax-M2.7"] | ||
|
|
||
| m3 = models["MiniMax-M3"] | ||
| assert m3["context_window"] == 1_000_000 | ||
| assert m3["max_output"] == 524_288 | ||
| assert m3["input_modalities"] == ["text", "image", "video"] | ||
| assert m3["thinking"] == ["adaptive", "disabled"] | ||
| assert m3["pricing_usd_per_million_tokens"] == { | ||
| "input": 0.6, | ||
| "output": 2.4, | ||
| "cache_read": 0.12, | ||
| "cache_write": None, | ||
| } | ||
| assert m3["pricing_tiers_usd_per_million_tokens"] == [ | ||
| { | ||
| "service_tier": "standard", | ||
| "input_tokens_lte": 512_000, | ||
| "input": 0.3, | ||
| "output": 1.2, | ||
| "cache_read": 0.06, | ||
| "cache_write": None, | ||
| }, | ||
| { | ||
| "service_tier": "standard", | ||
| "input_tokens_gt": 512_000, | ||
| "input": 0.6, | ||
| "output": 2.4, | ||
| "cache_read": 0.12, | ||
| "cache_write": None, | ||
| }, | ||
| { | ||
| "service_tier": "priority", | ||
| "input_tokens_lte": 512_000, | ||
| "input": 0.45, | ||
| "output": 1.8, | ||
| "cache_read": 0.09, | ||
| "cache_write": None, | ||
| }, | ||
| { | ||
| "service_tier": "priority", | ||
| "input_tokens_gt": 512_000, | ||
| "input": 0.9, | ||
| "output": 3.6, | ||
| "cache_read": 0.18, | ||
| "cache_write": None, | ||
| }, | ||
| ] | ||
|
|
||
| m27 = models["MiniMax-M2.7"] | ||
| assert m27["context_window"] == 204_800 | ||
| assert m27["max_output"] == 204_800 | ||
| assert m27["input_modalities"] == ["text"] | ||
| assert m27["thinking"] == ["always_on"] | ||
| assert m27["pricing_usd_per_million_tokens"] == { | ||
| "input": 0.3, | ||
| "output": 1.2, | ||
| "cache_read": 0.06, | ||
| "cache_write": 0.375, | ||
| } | ||
| PY |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
file="tests/test-minimax-model-catalog.sh"
block="$(sed -n '17,106p' "$file")"
grep -Fq 'assert provider["latest_planning"]' <<<"$block"
if grep -Eq '^[[:space:]]*if[[:space:]]+python3[[:space:]]+-E[[:space:]]+-' <<<"$block"; then
echo "Python environment isolation is present."
else
echo "Python assertions can be disabled by PYTHONOPTIMIZE."
exit 1
fiRepository: asklokesh/loki-mode
Length of output: 210
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
python3 - <<'PY'
import subprocess
import sys
import tempfile
import os
snippet = r'''
import json
from pathlib import Path
with open(Path(sys.argv[1]), encoding="utf-8") as handle:
catalog = json.load(handle)
provider = catalog["providers"]["minimax"]
assert provider["latest_planning"] == "MiniMax-M3"
assert provider["latest_development"] == "MiniMax-M3"
print("OK", provider["latest_planning"])
'''
catalog = '{"providers":{"minimax":{"latest_planning":"MiniMax-M42","latest_development":"MiniMax-M42","endpoints":[],"models":[{"id":"MiniMax-M3","context_window":999}]}}}'
with tempfile.NamedTemporaryFile("w", delete=False, encoding="utf-8") as fh:
fh.write(snippet)
py_path = fh.name
with tempfile.NamedTemporaryFile("w", delete=False, encoding="utf-8") as fh:
fh.write(catalog)
json_path = fh.name
for env in [{}, {"PYTHONOPTIMIZE": "", "PYTHONDONTWRITEBYTECODE": "1"}, {"PYTHONOPTIMIZE": "2", "PYTHONDONTWRITEBYTECODE": "1"}]:
env = dict(os.environ, **{k:v for k,v in env.items() if k == "PYTHONOPTIMIZE" or k == "PYTHONDONTWRITEBYTECODE" or k not in env})
result = subprocess.run([sys.executable, "-E", py_path, json_path], capture_output=True, text=True, env=env)
print("no flag", env.get("PYTHONOPTIMIZE"), "stdout=", result.stdout.strip(), "stderr=", result.stderr.strip(), "rc=", result.returncode)
env = dict(os.environ, **{k:v for k,v in env.items() if k == "PYTHONOPTIMIZE" or k == "PYTHONDONTWRITEBYTECODE" or k not in env})
result = subprocess.run([sys.executable, "-E", "-O", py_path, json_path], capture_output=True, text=True, env=env)
print("-O", env.get("PYTHONOPTIMIZE"), "stdout=", result.stdout.strip(), "stderr=", result.stderr.strip(), "rc=", result.returncode)
PYRepository: asklokesh/loki-mode
Length of output: 1430
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
python3 - <<'PY'
import subprocess
import sys
import tempfile
import os
snippet = r'''
import json
import sys
from pathlib import Path
with open(Path(sys.argv[1]), encoding="utf-8") as handle:
catalog = json.load(handle)
provider = catalog["providers"]["minimax"]
assert provider["latest_planning"] == "MiniMax-M3"
assert provider["latest_development"] == "MiniMax-M3"
print("OK", provider["latest_planning"])
'''
catalog = '{"providers":{"minimax":{"latest_planning":"MiniMax-M42","latest_development":"MiniMax-M42","endpoints":[],"models":[{"id":"MiniMax-M3","context_window":999}]}}}'
with tempfile.NamedTemporaryFile("w", delete=False, encoding="utf-8") as fh:
fh.write(snippet)
py_path = fh.name
with tempfile.NamedTemporaryFile("w", delete=False, encoding="utf-8") as fh:
fh.write(catalog)
json_path = fh.name
print("python", sys.version.split()[0])
for optimize in [None, 0, 1, 2]:
env = dict(os.environ)
if optimize is None:
env.pop("PYTHONOPTIMIZE", None)
else:
env["PYTHONOPTIMIZE"] = str(optimize)
for args in ([py_path, json_path], ["-O", py_path, json_path]):
result = subprocess.run([sys.executable] + args, capture_output=True, text=True, env=env)
print("env_PyOPTIMIZE=", optimize, "args=", args, "rc=", result.returncode, "stdout=", result.stdout.strip(), "stderr=", result.stderr.strip())
PYRepository: asklokesh/loki-mode
Length of output: 1299
Prevent optimized Python from disabling catalog checks.
PYTHONOPTIMIZE removes these assert checks, so the catalog test can pass after only parsing JSON. Use -E before invoking the validation script so the interpreter ignores environment-based optimization.
Proposed fix
-if python3 - "$CATALOG" <<'PY'
+if python3 -E - "$CATALOG" <<'PY'📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| if python3 - "$CATALOG" <<'PY' | |
| import json | |
| import sys | |
| with open(sys.argv[1], encoding="utf-8") as handle: | |
| catalog = json.load(handle) | |
| provider = catalog["providers"]["minimax"] | |
| assert provider["latest_planning"] == "MiniMax-M3" | |
| assert provider["latest_development"] == "MiniMax-M3" | |
| assert provider["latest_fast"] == "MiniMax-M3" | |
| endpoints = {item["region"]: item for item in provider["endpoints"]} | |
| assert endpoints == { | |
| "global_en": { | |
| "region": "global_en", | |
| "openai_base_url": "https://api.minimax.io/v1", | |
| "anthropic_base_url": "https://api.minimax.io/anthropic", | |
| "docs_root": "https://platform.minimax.io/docs", | |
| }, | |
| "cn_zh": { | |
| "region": "cn_zh", | |
| "openai_base_url": "https://api.minimaxi.com/v1", | |
| "anthropic_base_url": "https://api.minimaxi.com/anthropic", | |
| "docs_root": "https://platform.minimaxi.com/docs", | |
| }, | |
| } | |
| assert all(item["anthropic_base_url"].endswith("/anthropic") for item in endpoints.values()) | |
| models = {item["id"]: item for item in provider["models"]} | |
| assert list(models)[:2] == ["MiniMax-M3", "MiniMax-M2.7"] | |
| m3 = models["MiniMax-M3"] | |
| assert m3["context_window"] == 1_000_000 | |
| assert m3["max_output"] == 524_288 | |
| assert m3["input_modalities"] == ["text", "image", "video"] | |
| assert m3["thinking"] == ["adaptive", "disabled"] | |
| assert m3["pricing_usd_per_million_tokens"] == { | |
| "input": 0.6, | |
| "output": 2.4, | |
| "cache_read": 0.12, | |
| "cache_write": None, | |
| } | |
| assert m3["pricing_tiers_usd_per_million_tokens"] == [ | |
| { | |
| "service_tier": "standard", | |
| "input_tokens_lte": 512_000, | |
| "input": 0.3, | |
| "output": 1.2, | |
| "cache_read": 0.06, | |
| "cache_write": None, | |
| }, | |
| { | |
| "service_tier": "standard", | |
| "input_tokens_gt": 512_000, | |
| "input": 0.6, | |
| "output": 2.4, | |
| "cache_read": 0.12, | |
| "cache_write": None, | |
| }, | |
| { | |
| "service_tier": "priority", | |
| "input_tokens_lte": 512_000, | |
| "input": 0.45, | |
| "output": 1.8, | |
| "cache_read": 0.09, | |
| "cache_write": None, | |
| }, | |
| { | |
| "service_tier": "priority", | |
| "input_tokens_gt": 512_000, | |
| "input": 0.9, | |
| "output": 3.6, | |
| "cache_read": 0.18, | |
| "cache_write": None, | |
| }, | |
| ] | |
| m27 = models["MiniMax-M2.7"] | |
| assert m27["context_window"] == 204_800 | |
| assert m27["max_output"] == 204_800 | |
| assert m27["input_modalities"] == ["text"] | |
| assert m27["thinking"] == ["always_on"] | |
| assert m27["pricing_usd_per_million_tokens"] == { | |
| "input": 0.3, | |
| "output": 1.2, | |
| "cache_read": 0.06, | |
| "cache_write": 0.375, | |
| } | |
| PY | |
| if python3 -E - "$CATALOG" <<'PY' | |
| import json | |
| import sys | |
| with open(sys.argv[1], encoding="utf-8") as handle: | |
| catalog = json.load(handle) | |
| provider = catalog["providers"]["minimax"] | |
| assert provider["latest_planning"] == "MiniMax-M3" | |
| assert provider["latest_development"] == "MiniMax-M3" | |
| assert provider["latest_fast"] == "MiniMax-M3" | |
| endpoints = {item["region"]: item for item in provider["endpoints"]} | |
| assert endpoints == { | |
| "global_en": { | |
| "region": "global_en", | |
| "openai_base_url": "https://api.minimax.io/v1", | |
| "anthropic_base_url": "https://api.minimax.io/anthropic", | |
| "docs_root": "https://platform.minimax.io/docs", | |
| }, | |
| "cn_zh": { | |
| "region": "cn_zh", | |
| "openai_base_url": "https://api.minimaxi.com/v1", | |
| "anthropic_base_url": "https://api.minimaxi.com/anthropic", | |
| "docs_root": "https://platform.minimaxi.com/docs", | |
| }, | |
| } | |
| assert all(item["anthropic_base_url"].endswith("/anthropic") for item in endpoints.values()) | |
| models = {item["id"]: item for item in provider["models"]} | |
| assert list(models)[:2] == ["MiniMax-M3", "MiniMax-M2.7"] | |
| m3 = models["MiniMax-M3"] | |
| assert m3["context_window"] == 1_000_000 | |
| assert m3["max_output"] == 524_288 | |
| assert m3["input_modalities"] == ["text", "image", "video"] | |
| assert m3["thinking"] == ["adaptive", "disabled"] | |
| assert m3["pricing_usd_per_million_tokens"] == { | |
| "input": 0.6, | |
| "output": 2.4, | |
| "cache_read": 0.12, | |
| "cache_write": None, | |
| } | |
| assert m3["pricing_tiers_usd_per_million_tokens"] == [ | |
| { | |
| "service_tier": "standard", | |
| "input_tokens_lte": 512_000, | |
| "input": 0.3, | |
| "output": 1.2, | |
| "cache_read": 0.06, | |
| "cache_write": None, | |
| }, | |
| { | |
| "service_tier": "standard", | |
| "input_tokens_gt": 512_000, | |
| "input": 0.6, | |
| "output": 2.4, | |
| "cache_read": 0.12, | |
| "cache_write": None, | |
| }, | |
| { | |
| "service_tier": "priority", | |
| "input_tokens_lte": 512_000, | |
| "input": 0.45, | |
| "output": 1.8, | |
| "cache_read": 0.09, | |
| "cache_write": None, | |
| }, | |
| { | |
| "service_tier": "priority", | |
| "input_tokens_gt": 512_000, | |
| "input": 0.9, | |
| "output": 3.6, | |
| "cache_read": 0.18, | |
| "cache_write": None, | |
| }, | |
| ] | |
| m27 = models["MiniMax-M2.7"] | |
| assert m27["context_window"] == 204_800 | |
| assert m27["max_output"] == 204_800 | |
| assert m27["input_modalities"] == ["text"] | |
| assert m27["thinking"] == ["always_on"] | |
| assert m27["pricing_usd_per_million_tokens"] == { | |
| "input": 0.3, | |
| "output": 1.2, | |
| "cache_read": 0.06, | |
| "cache_write": 0.375, | |
| } | |
| PY |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tests/test-minimax-model-catalog.sh` around lines 17 - 106, Update the Python
invocation wrapping the catalog validation script to include the -E interpreter
option before executing the heredoc, ensuring environment-based optimization
cannot remove the assert checks in the validation code. Keep the existing
catalog parsing and assertions unchanged.
|
Thanks for the review. I resolved the merge conflict against current main and pushed commit 7bf021e. Targeted MiniMax catalog, single-source, flagship-consistency, provider-scoping, JSON, and shell-syntax checks passed; the generic-tier suite is blocked locally by macOS Bash 3.2 lacking mapfile. |
1 participant
Summary
Verification
python3 -m json.tool providers/model_catalog.jsonbash -n tests/test-minimax-model-catalog.shbash -n tests/run-all-tests.shbash tests/test-minimax-model-catalog.shbash tests/test-cli-provider-flag.shbash tests/test-model-and-port.shgit diff --check