effgen 0.2.2__tar.gz → 0.2.3__tar.gz

{effgen-0.2.2/effgen.egg-info → effgen-0.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: effgen
-Version: 0.2.2
+Version: 0.2.3
 Summary: A comprehensive framework for building agents with Small Language Models
 Home-page: https://github.com/ctrl-gaurav/effGen
 Author: Gaurav Srivastava
@@ -108,6 +108,16 @@ Provides-Extra: gguf
 Requires-Dist: llama-cpp-python>=0.2.0; extra == "gguf"
 Provides-Extra: cerebras
 Requires-Dist: cerebras-cloud-sdk>=1.0; extra == "cerebras"
+Provides-Extra: groq
+Requires-Dist: groq>=0.15; extra == "groq"
+Provides-Extra: together
+Requires-Dist: together>=1.3; extra == "together"
+Provides-Extra: fireworks
+Requires-Dist: fireworks-ai>=0.15; extra == "fireworks"
+Provides-Extra: replicate
+Requires-Dist: replicate>=1.0; extra == "replicate"
+Provides-Extra: hf
+Requires-Dist: huggingface_hub>=0.26; extra == "hf"
 Provides-Extra: flash-attn
 Requires-Dist: flash-attn>=2.3.0; extra == "flash-attn"
 Provides-Extra: vector-db
@@ -148,6 +158,11 @@ Requires-Dist: rouge-score>=0.1.2; extra == "all"
 Requires-Dist: nltk>=3.8.0; extra == "all"
 Requires-Dist: llama-cpp-python>=0.2.0; extra == "all"
 Requires-Dist: cerebras-cloud-sdk>=1.0; extra == "all"
+Requires-Dist: groq>=0.15; extra == "all"
+Requires-Dist: together>=1.3; extra == "all"
+Requires-Dist: fireworks-ai>=0.15; extra == "all"
+Requires-Dist: replicate>=1.0; extra == "all"
+Requires-Dist: huggingface_hub>=0.26; extra == "all"
 Requires-Dist: bitsandbytes>=0.46.1; extra == "all"
 Requires-Dist: datasets>=2.14.0; extra == "all"
 Dynamic: author

{effgen-0.2.2 → effgen-0.2.3}/README.md

@@ -36,6 +36,7 @@
 
 | | Date | Update |
 |:---:|:---|:---|
+| 🚀 | **4 May 2026** | **v0.2.3 Released**: 5 new cloud backends (Groq, Together AI, Fireworks, Replicate, HuggingFace Inference) — 9 providers total. Unified ProviderRegistry, `effgen doctor` auth check, backend parity matrix. [See changelog](CHANGELOG.md#023---2026-05-04) |
 | 🚀 | **28 Apr 2026** | **v0.2.2 Released**: Gemini 3.x/2.5/2.0 registry, `thinking_budget`, Google Search grounding, Files API, Gemini native tools (GoogleSearch, UrlContext, CodeExecution). Anthropic Claude 4.7 registry, extended thinking, prompt caching (`cache_control`), streaming polish, experimental native tools. [See changelog](CHANGELOG.md#022---2026-04-28) |
 | 🚀 | **25 Apr 2026** | **v0.2.1 Released**: Cerebras backend (4 free-tier models, streaming, native tool-calling, rate-limit coordinator, cost tracking) + OpenAI gpt-5/gpt-5.4-nano/o-series with `reasoning_effort`, prompt caching, structured outputs v2, and OpenAI native tools (web_search, code_interpreter, file_search). [See changelog](CHANGELOG.md#021---2026-04-25) |
 | 🚀 | **9 Apr 2026** | **v0.2.0 Released**: Major release — native tool calling, guardrails, multi-agent orchestration, RAG pipeline, 31 tools, eval framework, production API server, MLX Apple Silicon support, Python & TypeScript SDKs. [See changelog](CHANGELOG.md#020---2026-04-09) |
@@ -270,10 +271,30 @@ Production API<br/>
 
 ---
 
-## 🆕 What's New in v0.2.2
+## 🆕 What's New in v0.2.3
 
 <details open>
-<summary><b>Top 5 features in v0.2.2</b></summary>
+<summary><b>Top 5 features in v0.2.3</b></summary>
+
+1. **5 new cloud backends** — `GroqAdapter`, `TogetherAdapter`, `FireworksAdapter`, `ReplicateAdapter`, `HFInferenceAdapter` — each with streaming, native tools, rate-limit coordination, and cost tracking. 9 providers total.
+
+   ```python
+   model = load_model("llama-3.1-8b-instant", provider="groq")
+   model = load_model("Qwen/Qwen2.5-72B-Instruct", provider="hf")
+   ```
+
+2. **Unified ProviderRegistry** — `list_providers()`, `list_models(provider)`, `lookup(model_id)` consolidated across all 9 adapters. `AmbiguousModelError` on bare IDs shared across providers.
+
+3. **`effgen doctor`** — new CLI command showing which providers have API keys configured.
+
+4. **Backend parity matrix** — canonical agentic task ("(17 × 23) + sqrt(144) = 403") runs identically across all providers; streaming and error surfaces verified uniform. See `docs/providers/parity.md`.
+
+5. **HuggingFace Router support** — `HFInferenceAdapter` with 124-model dynamic catalog, `refresh_models()` + `check_drift()`, `ModelUnavailableError` with `suggest_alternatives()`, and custom Inference Endpoint URL.
+
+</details>
+
+<details>
+<summary><b>Top 5 features from v0.2.2</b></summary>
 
 1. **Gemini 3.x/2.5/2.0 + Gemma families** — full model registry with correct context windows, output limits, and feature flags; SDK migrated to `google-genai>=1.0.0`.
 
@@ -586,15 +607,47 @@ result = agent.run("What does the documentation say about configuration?")
 
 ## 🤖 Multi-Model Support
 
-effGen supports **7 inference backends** and is tested across 11+ model families:
+effGen supports **9 cloud inference providers** + 4 local backends, tested across 11+ model families:
+
+| Backend | Platform | Install | Best For |
+|---------|----------|---------|----------|
+| **MLX** | Apple Silicon (M1/M2/M3/M4) | `effgen[mlx]` | Native Metal GPU, unified memory, 4/8-bit quantization |
+| **MLX-VLM** | Apple Silicon | `effgen[mlx-vlm]` | Vision-Language models (Qwen2-VL, LLaVA, Phi-3 Vision, 30+ architectures) |
+| **vLLM** | NVIDIA GPU | `effgen[vllm]` | High-throughput batch inference |
+| **Transformers** | Any (CPU/GPU) | *(bundled)* | Universal compatibility, local models |
+| **OpenAI** | Cloud API | *(bundled)* | gpt-5/gpt-5.4/o-series, reasoning_effort, structured outputs, native tools |
+| **Anthropic** | Cloud API | *(bundled)* | Claude 4.7/4.x, extended thinking, prompt caching, native tools |
+| **Google Gemini** | Cloud API | *(bundled)* | Gemini 3.x/2.5/2.0, thinking_budget, grounding, Files API, native tools |
+| **Cerebras** | Cloud API | `effgen[cerebras]` | 4 free-tier models (llama3.1-8b, qwen-3-235b), ultra-low latency |
+| **Groq** | Cloud API | `effgen[groq]` | 16 models (llama-3.3-70b, mixtral, qwen3-32b), ultra-fast free-tier inference |
+| **Together AI** | Cloud API | `effgen[together]` | 163-model catalog (llama, deepseek, qwen, mistral), per-model pricing |
+| **Fireworks** | Cloud API | `effgen[fireworks]` | 80 chat models (54 tool-capable), serverless + dedicated |
+| **Replicate** | Cloud API | `effgen[replicate]` | 38 models, async run-poll, SSE streaming, compute-second billing |
+| **HuggingFace** | Cloud API | `effgen[hf]` | 124-model HF Router catalog, custom Inference Endpoints, free serverless tier |
+
+### Provider Auth Check
+
+```bash
+# See which API keys are configured
+effgen doctor
+```
+
+### Quick Cloud Start
 
-| Backend | Platform | Best For |
-|---------|----------|----------|
-| **MLX** | Apple Silicon (M1/M2/M3/M4) | Native Metal GPU, unified memory, 4/8-bit quantization |
-| **MLX-VLM** | Apple Silicon | Vision-Language models (Qwen2-VL, LLaVA, Phi-3 Vision, 30+ architectures) |
-| **vLLM** | NVIDIA GPU | High-throughput batch inference |
-| **Transformers** | Any (CPU/GPU) | Universal compatibility |
-| **API** | Cloud | OpenAI (gpt-5/gpt-5.4/o-series + reasoning_effort), Anthropic (Claude 4.7/4.x + thinking + caching), Google Gemini (3.x/2.5/2.0 + thinking_budget + grounding + Files API + native tools), Cerebras (4 free-tier models, streaming + native tools) |
+```python
+from effgen import load_model, Agent
+from effgen.core.agent import AgentConfig
+from effgen.tools.builtin import Calculator
+
+# Any of the 9 cloud providers
+model = load_model("llama-3.1-8b-instant", provider="groq")          # Groq
+# model = load_model("meta-llama/Llama-3.3-70B-Instruct-Turbo", provider="together")
+# model = load_model("Qwen/Qwen2.5-72B-Instruct", provider="hf")
+
+agent = Agent(config=AgentConfig(name="agent", model=model, tools=[Calculator()]))
+result = agent.run("What is (17 * 23) + sqrt(144)?")
+print(result.output)  # → 403
+```
 
 ### Top Recommended Models
 

{effgen-0.2.2 → effgen-0.2.3}/effgen/__init__.py

@@ -9,7 +9,7 @@ This framework enables SLMs to function as powerful agentic systems through:
 - Comprehensive configuration management
 """
 
-__version__ = "0.2.2"
+__version__ = "0.2.3"
 __author__ = "effGen Team"
 __license__ = "Apache-2.0"
 
@@ -74,30 +74,75 @@ from effgen.models import (
     AnthropicAdapter,
     BaseModel,
     CerebrasAdapter,
+    FireworksAdapter,
     GeminiAdapter,
     GenerationConfig,
     GenerationResult,
+    GroqAdapter,
+    HFInferenceAdapter,
     ModelLoader,
     OpenAIAdapter,
+    ReplicateAdapter,
     StreamChunk,
+    TogetherAdapter,
     TransformersEngine,
     VLLMEngine,
     load_model,
 )
 from effgen.models._rate_limit import RateLimitCoordinator, RateLimitExceeded  # noqa: I001
+from effgen.models.auth import check_keys
 from effgen.models.cerebras_models import available_models as cerebras_available_models
 from effgen.models.cerebras_models import free_tier_models as cerebras_free_tier_models
 from effgen.models.cerebras_models import model_info as cerebras_model_info
-from effgen.models.errors import ModelRefusalError, ToolIncompatibleError
+from effgen.models.errors import (
+    AmbiguousModelError,
+    ModelAuthError,
+    ModelNotFoundError,
+    ModelRefusalError,
+    ModelTimeoutError,
+    ModelUnavailableError,
+    ToolIncompatibleError,
+)
+from effgen.models.fireworks_models import available_models as fireworks_available_models
+from effgen.models.fireworks_models import chat_models as fireworks_chat_models
+from effgen.models.fireworks_models import pricing_table as fireworks_pricing_table
+from effgen.models.fireworks_models import refresh_models as fireworks_refresh_models
+from effgen.models.fireworks_models import tool_capable_models as fireworks_tool_capable_models
 from effgen.models.gemini_models import available_models as gemini_available_models
 from effgen.models.gemini_models import free_tier_models as gemini_free_tier_models
 from effgen.models.gemini_models import model_info as gemini_model_info
 from effgen.models.gemini_models import recommended_models as gemini_recommended_models
+from effgen.models.groq_models import available_models as groq_available_models
+from effgen.models.groq_models import chat_models as groq_chat_models
+from effgen.models.groq_models import tool_capable_models as groq_tool_capable_models
+from effgen.models.hf_inference_models import available_models as hf_available_models
+from effgen.models.hf_inference_models import catalog_summary as hf_catalog_summary
+from effgen.models.hf_inference_models import chat_models as hf_chat_models
+from effgen.models.hf_inference_models import cheapest_provider as hf_cheapest_provider
+from effgen.models.hf_inference_models import check_drift as hf_check_drift
+from effgen.models.hf_inference_models import get_model_info as hf_get_model_info
+from effgen.models.hf_inference_models import list_providers_for as hf_list_providers_for
+from effgen.models.hf_inference_models import refresh_models as hf_refresh_models
+from effgen.models.hf_inference_models import serverless_models as hf_serverless_models
+from effgen.models.hf_inference_models import suggest_alternatives as hf_suggest_alternatives
+from effgen.models.hf_inference_models import tool_capable_models as hf_tool_capable_models
 from effgen.models.openai_models import available_models as openai_available_models
 from effgen.models.openai_models import chat_models as openai_chat_models
 from effgen.models.openai_models import model_info as openai_model_info
 from effgen.models.openai_models import reasoning_models as openai_reasoning_models  # noqa: I001
 from effgen.models.openai_schema import to_openai_schema
+from effgen.models.registry import ProviderRegistry, list_models, list_providers, lookup
+from effgen.models.replicate_models import available_models as replicate_available_models
+from effgen.models.replicate_models import get_model_info as replicate_get_model_info
+from effgen.models.replicate_models import refresh_models as replicate_refresh_models
+from effgen.models.replicate_models import streaming_models as replicate_streaming_models
+from effgen.models.replicate_models import tool_capable_models as replicate_tool_capable_models
+from effgen.models.together_models import available_models as together_available_models
+from effgen.models.together_models import chat_models as together_chat_models
+from effgen.models.together_models import pricing_table as together_pricing_table
+from effgen.models.together_models import refresh_models as together_refresh_models
+from effgen.models.together_models import serverless_models as together_serverless_models
+from effgen.models.together_models import tool_capable_models as together_tool_capable_models
 
 # Preset imports
 from effgen.presets import create_agent, list_presets
@@ -204,17 +249,38 @@ __all__ = [
     "StreamChunk",
     "GeminiAdapter",
     "CerebrasAdapter",
+    "GroqAdapter",
+    "TogetherAdapter",
+    "FireworksAdapter",
+    "ReplicateAdapter",
+    "HFInferenceAdapter",
     "ModelLoader",
     "GenerationConfig",
     "GenerationResult",
     "RateLimitCoordinator",
     "RateLimitExceeded",
     "ModelRefusalError",
+    "ModelAuthError",
+    "ModelTimeoutError",
+    "ModelUnavailableError",
+    "ModelNotFoundError",
+    "AmbiguousModelError",
+    "ToolIncompatibleError",
     "to_openai_schema",
+    # Provider registry + auth
+    "ProviderRegistry",
+    "list_providers",
+    "list_models",
+    "lookup",
+    "check_keys",
     # Cerebras helpers
     "cerebras_available_models",
     "cerebras_free_tier_models",
     "cerebras_model_info",
+    # Groq helpers
+    "groq_available_models",
+    "groq_chat_models",
+    "groq_tool_capable_models",
     # OpenAI helpers
     "openai_available_models",
     "openai_chat_models",
@@ -225,6 +291,37 @@ __all__ = [
     "gemini_free_tier_models",
     "gemini_model_info",
     "gemini_recommended_models",
+    # Together helpers
+    "together_available_models",
+    "together_chat_models",
+    "together_tool_capable_models",
+    "together_pricing_table",
+    "together_refresh_models",
+    "together_serverless_models",
+    # Fireworks helpers
+    "fireworks_available_models",
+    "fireworks_chat_models",
+    "fireworks_tool_capable_models",
+    "fireworks_pricing_table",
+    "fireworks_refresh_models",
+    # Replicate helpers
+    "replicate_available_models",
+    "replicate_streaming_models",
+    "replicate_tool_capable_models",
+    "replicate_refresh_models",
+    "replicate_get_model_info",
+    # HF Inference helpers
+    "hf_available_models",
+    "hf_chat_models",
+    "hf_tool_capable_models",
+    "hf_serverless_models",
+    "hf_suggest_alternatives",
+    "hf_get_model_info",
+    "hf_refresh_models",
+    "hf_check_drift",
+    "hf_catalog_summary",
+    "hf_list_providers_for",
+    "hf_cheapest_provider",
 
     # Tools
     "BaseTool",

{effgen-0.2.2 → effgen-0.2.3}/effgen/cli.py

@@ -1832,6 +1832,13 @@ Examples:
     # Health check command
     subparsers.add_parser('health', help='Check effGen infrastructure health')
 
+    # Doctor command — API key availability check
+    doctor_parser = subparsers.add_parser('doctor', help='Check provider API key availability')
+    doctor_parser.add_argument('--json', dest='output_json', action='store_true',
+                               help='Output as JSON')
+    doctor_parser.add_argument('--provider', dest='doctor_provider',
+                               help='Check a specific provider only')
+
     # Plugin commands
     plugin_parser = subparsers.add_parser('create-plugin', help='Generate a plugin project scaffold')
     plugin_parser.add_argument('plugin_name', help='Plugin name (e.g. my_tools)')
@@ -2007,6 +2014,101 @@ dependencies = ["effgen"]
     return 0
 
 
+def _handle_doctor_command(args) -> int:
+    """Handle the 'effgen doctor' subcommand — check API key availability."""
+    import json as _json
+
+    # Load .env from standard locations before checking keys (all, non-overriding)
+    try:
+        from dotenv import load_dotenv
+        for _env_path in [
+            Path.home() / ".effgen" / ".env",
+            Path(".env"),
+            Path(__file__).parent.parent / ".env",
+        ]:
+            if _env_path.exists():
+                load_dotenv(_env_path, override=False)
+    except ImportError:
+        pass
+
+    from effgen.models.auth import check_keys
+    from effgen.models.registry import ProviderRegistry
+
+    # Ensure all adapters are imported so they self-register
+    try:
+        import effgen.models.anthropic_adapter  # noqa: F401
+        import effgen.models.cerebras_adapter  # noqa: F401
+        import effgen.models.fireworks_adapter  # noqa: F401
+        import effgen.models.gemini_adapter  # noqa: F401
+        import effgen.models.groq_adapter  # noqa: F401
+        import effgen.models.hf_inference_adapter  # noqa: F401
+        import effgen.models.openai_adapter  # noqa: F401
+        import effgen.models.replicate_adapter  # noqa: F401
+        import effgen.models.together_adapter  # noqa: F401
+    except Exception:
+        pass
+
+    provider_filter = getattr(args, 'doctor_provider', None)
+    providers_to_check = [provider_filter] if provider_filter else None
+
+    results = check_keys(providers_to_check)
+
+    if getattr(args, 'output_json', False):
+        print(_json.dumps(results, indent=2))
+        return 0
+
+    # Pretty-print
+    if RICH_AVAILABLE:
+        console = Console()
+        table = Table(title="effgen doctor — Provider API Key Status")
+        table.add_column("Provider", style="cyan", no_wrap=True)
+        table.add_column("Status", style="white")
+        table.add_column("Env Key Found", style="dim")
+        table.add_column("Models", style="dim", justify="right")
+
+        for prov in sorted(results):
+            info = results[prov]
+            available = info.get("available", False)
+            env_key = info.get("env_key") or "—"
+            status = "[green]READY[/green]" if available else "[red]MISSING KEY[/red]"
+            try:
+                n_models = str(len(ProviderRegistry.list_models(prov)))
+            except Exception:
+                n_models = "?"
+            table.add_row(prov, status, env_key, n_models)
+
+        console.print(table)
+
+        # Print hints for missing keys
+        missing = [p for p, i in results.items() if not i.get("available")]
+        if missing:
+            console.print("\n[yellow]Missing keys — set in ~/.effgen/.env or export:[/yellow]")
+            for prov in missing:
+                keys = results[prov].get("env_keys_checked", [])
+                key_str = " or ".join(keys) if keys else f"{prov.upper()}_API_KEY"
+                console.print(f"  export {key_str}=<your-key>")
+        else:
+            console.print("\n[green]All providers ready![/green]")
+    else:
+        print("effgen doctor — Provider API Key Status")
+        print("-" * 50)
+        for prov in sorted(results):
+            info = results[prov]
+            available = info.get("available", False)
+            env_key = info.get("env_key") or "not set"
+            status = "READY" if available else "MISSING KEY"
+            print(f"  {prov:15s} {status:12s}  (key: {env_key})")
+        missing = [p for p, i in results.items() if not i.get("available")]
+        if missing:
+            print("\nMissing keys — set in ~/.effgen/.env or export:")
+            for prov in missing:
+                keys = results[prov].get("env_keys_checked", [])
+                key_str = " or ".join(keys) if keys else f"{prov.upper()}_API_KEY"
+                print(f"  export {key_str}=<your-key>")
+
+    return 0
+
+
 def _handle_workflow_command(args, cli) -> int:
     """Handle the 'workflow' CLI subcommand."""
     from effgen.core.workflow import WorkflowDAG
@@ -2100,7 +2202,7 @@ def _handle_batch_command(args, cli) -> int:
             config = AgentConfig(name="batch-agent", model=model, max_iterations=5)
             agent = Agent(config)
 
-        config = BatchConfig(
+        batch_config = BatchConfig(
             max_concurrency=args.concurrency,
             batch_size=args.batch_size,
             retry_failed=args.retries,
@@ -2109,7 +2211,7 @@ def _handle_batch_command(args, cli) -> int:
 
         runner = BatchRunner(agent)
         cli.print(f"Loading queries from {input_path}...")
-        result = runner.run_from_file(input_path, config=config, query_field=query_field)
+        result = runner.run_from_file(input_path, config=batch_config, query_field=query_field)
 
         cli.print(
             f"\nBatch complete: {result.succeeded}/{result.total} succeeded "
@@ -2405,6 +2507,8 @@ def main():
             checker = HealthChecker()
             all_passed = checker.print_results()
             exit_code = 0 if all_passed else 1
+        elif args.command == 'doctor':
+            exit_code = _handle_doctor_command(args)
         elif args.command == 'resume':
             exit_code = _handle_resume_command(args, cli)
         elif args.command == 'sessions':

{effgen-0.2.2 → effgen-0.2.3}/effgen/models/__init__.py

@@ -33,6 +33,7 @@ from effgen.models.anthropic_cache import (
 )
 from effgen.models.anthropic_models import ANTHROPIC_MODELS
 from effgen.models.anthropic_models import get_model_info as get_anthropic_model_info
+from effgen.models.auth import check_keys
 from effgen.models.base import (
     BaseModel,
     BatchModel,
@@ -52,14 +53,30 @@ from effgen.models.capabilities import (
     register_model_capability,
 )
 from effgen.models.cerebras_adapter import CerebrasAdapter
-from effgen.models.errors import ModelRefusalError
+from effgen.models.errors import (
+    AmbiguousModelError,
+    ModelAuthError,
+    ModelNotFoundError,
+    ModelRefusalError,
+    ModelTimeoutError,
+    ModelUnavailableError,
+)
+from effgen.models.fireworks_adapter import FireworksAdapter
+from effgen.models.fireworks_models import FIREWORKS_MODELS
 from effgen.models.gemini_adapter import GeminiAdapter
 from effgen.models.gemini_files import FileRef, upload_file
+from effgen.models.groq_adapter import GroqAdapter
+from effgen.models.groq_models import GROQ_MODELS
+from effgen.models.hf_inference_adapter import HFInferenceAdapter
+from effgen.models.hf_inference_models import HF_MODELS
 from effgen.models.lazy import LazyModel
 from effgen.models.model_loader import ModelLoader, load_model
 from effgen.models.openai_adapter import OpenAIAdapter
 from effgen.models.openai_schema import to_openai_schema
 from effgen.models.pool import ModelPool, PoolConfig
+from effgen.models.registry import ProviderRegistry, list_models, list_providers, lookup
+from effgen.models.replicate_adapter import ReplicateAdapter
+from effgen.models.replicate_models import REPLICATE_MODELS
 from effgen.models.router import (
     ComplexityEstimate,
     ComplexityLevel,
@@ -68,6 +85,8 @@ from effgen.models.router import (
     RoutingDecision,
     estimate_complexity,
 )
+from effgen.models.together_adapter import TogetherAdapter
+from effgen.models.together_models import TOGETHER_MODELS
 from effgen.models.transformers_engine import TransformersEngine
 from effgen.models.vllm_engine import VLLMEngine
 
@@ -105,6 +124,18 @@ __all__ = [
     "AnthropicAdapter",
     "GeminiAdapter",
     "CerebrasAdapter",
+    "GroqAdapter",
+    "TogetherAdapter",
+    "FireworksAdapter",
+    "ReplicateAdapter",
+    "HFInferenceAdapter",
+
+    # Model registries
+    "GROQ_MODELS",
+    "TOGETHER_MODELS",
+    "FIREWORKS_MODELS",
+    "REPLICATE_MODELS",
+    "HF_MODELS",
 
     # Anthropic streaming
     "StreamChunk",
@@ -138,6 +169,18 @@ __all__ = [
 
     # Errors
     "ModelRefusalError",
+    "ModelAuthError",
+    "ModelTimeoutError",
+    "ModelUnavailableError",
+    "ModelNotFoundError",
+    "AmbiguousModelError",
+
+    # Provider registry + auth
+    "ProviderRegistry",
+    "list_providers",
+    "list_models",
+    "lookup",
+    "check_keys",
 
     # Schema helpers
     "to_openai_schema",

{effgen-0.2.2 → effgen-0.2.3}/effgen/models/_cost.py

@@ -63,11 +63,99 @@ _RATES: dict[str, dict[str, tuple[float, float]]] = {
         "gemini-1.5-pro": (3.50, 10.50),
         "*": (1.00, 3.00),
     },
+    "groq": {
+        # OFFICIAL: Groq free tier = $0 for all models (2026-04-28)
+        "*": (0.0, 0.0),
+    },
+    "together": {
+        # OFFICIAL rates from Together AI pricing page (2026-04-28).
+        # Per million tokens: (input, output)
+        "meta-llama/Llama-3.3-70B-Instruct-Turbo": (0.88, 0.88),
+        "meta-llama/Meta-Llama-3-8B-Instruct-Lite": (0.10, 0.10),
+        "meta-llama/Meta-Llama-3-8B-Instruct": (0.20, 0.20),
+        "meta-llama/Llama-4-Scout-17B-16E-Instruct": (0.18, 0.59),
+        "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8": (0.27, 0.85),
+        "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo": (0.18, 0.18),
+        "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo": (0.88, 0.88),
+        "meta-llama/Meta-Llama-3-70B-Instruct-Turbo": (0.88, 0.88),
+        "meta-llama/Meta-Llama-3.1-405B-Instruct": (3.50, 3.50),
+        "meta-llama/Llama-3.2-1B-Instruct": (0.06, 0.06),
+        "meta-llama/Llama-3-8b-chat-hf": (0.20, 0.20),
+        "Qwen/Qwen2.5-7B-Instruct-Turbo": (0.30, 0.30),
+        "Qwen/Qwen2.5-72B-Instruct-Turbo": (1.20, 1.20),
+        "Qwen/Qwen2.5-72B-Instruct": (1.20, 1.20),
+        "Qwen/Qwen2.5-14B-Instruct": (0.80, 0.80),
+        "Qwen/Qwen2.5-Coder-32B-Instruct": (0.80, 0.80),
+        "Qwen/QwQ-32B": (1.20, 1.20),
+        "Qwen/Qwen3.5-9B": (0.10, 0.15),
+        "Qwen/Qwen3.5-397B-A17B": (0.60, 3.60),
+        "Qwen/Qwen3-235B-A22B-Instruct-2507-tput": (0.20, 0.60),
+        "Qwen/Qwen3-Coder-Next-FP8": (0.50, 1.20),
+        "Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8": (2.00, 2.00),
+        "Qwen/Qwen3-235B-A22B-Thinking-2507": (0.65, 3.00),
+        "Qwen/Qwen3-Next-80B-A3B-Instruct": (0.15, 1.50),
+        "Qwen/Qwen3-Next-80B-A3B-Thinking": (0.15, 1.50),
+        "Qwen/Qwen3-VL-8B-Instruct": (0.18, 0.68),
+        "Qwen/Qwen3-VL-32B-Instruct": (0.50, 1.50),
+        "Qwen/Qwen2-VL-72B-Instruct": (1.20, 1.20),
+        "Qwen/Qwen2.5-VL-72B-Instruct": (1.95, 8.00),
+        "deepseek-ai/DeepSeek-V3.1": (0.60, 1.70),
+        "deepseek-ai/DeepSeek-V3-0324": (1.25, 1.25),
+        "deepseek-ai/DeepSeek-V4-Pro": (2.10, 4.40),
+        "deepseek-ai/DeepSeek-R1": (3.00, 7.00),
+        "deepseek-ai/DeepSeek-R1-0528": (3.00, 7.00),
+        "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B": (0.18, 0.18),
+        "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B": (0.00, 0.00),
+        "deepseek-ai/DeepSeek-R1-Distill-Qwen-14B": (1.60, 1.60),
+        "deepseek-ai/DeepSeek-R1-Distill-Llama-70B": (2.00, 2.00),
+        "mistralai/Mixtral-8x7B-Instruct-v0.1": (0.60, 0.60),
+        "mistralai/Mistral-7B-Instruct-v0.3": (0.20, 0.20),
+        "mistralai/Mistral-7B-Instruct-v0.1": (0.20, 0.20),
+        "mistralai/Mistral-Small-24B-Instruct-2501": (0.10, 0.30),
+        "mistralai/Ministral-3-14B-Instruct-2512": (0.20, 0.20),
+        "openai/gpt-oss-20b": (0.05, 0.20),
+        "openai/gpt-oss-120b": (0.15, 0.60),
+        "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF": (0.88, 0.88),
+        "nvidia/NVIDIA-Nemotron-Nano-9B-v2": (0.06, 0.25),
+        "moonshotai/Kimi-K2.5": (0.50, 2.80),
+        "moonshotai/Kimi-K2.6": (1.20, 4.50),
+        "moonshotai/Kimi-K2-Thinking": (1.20, 4.00),
+        "MiniMaxAI/MiniMax-M2.5": (0.30, 1.20),
+        "MiniMaxAI/MiniMax-M2.7": (0.30, 1.20),
+        "MiniMaxAI/MiniMax-M2": (0.00, 0.00),
+        "zai-org/GLM-4.5-Air-FP8": (0.20, 1.10),
+        "zai-org/GLM-4.6": (0.60, 2.20),
+        "zai-org/GLM-4.7": (0.45, 2.00),
+        "zai-org/GLM-5": (1.00, 3.20),
+        "zai-org/GLM-5.1": (1.40, 4.40),
+        "google/gemma-4-31B-it": (0.20, 0.50),
+        "google/gemma-3n-E4B-it": (0.06, 0.12),
+        "LiquidAI/LFM2-24B-A2B": (0.03, 0.12),
+        "arize-ai/qwen-2-1.5b-instruct": (0.10, 0.10),
+        "essentialai/rnj-1-instruct": (0.15, 0.15),
+        "deepcogito/cogito-v2-1-671b": (1.25, 1.25),
+        "Qwen/Qwen2-1.5B-Instruct": (0.02, 0.02),
+        # Free / dedicated-endpoint models → $0 in tracker
+        "*": (0.0, 0.0),
+    },
 }
 
 
 def _rate(provider: str, model: str) -> tuple[float, float]:
     """Lookup (input_per_M, output_per_M) rate for provider/model."""
+    if provider.lower() == "fireworks":
+        try:
+            from effgen.models.fireworks_models import FIREWORKS_MODELS
+            info = FIREWORKS_MODELS.get(model)
+            if info is not None:
+                return (
+                    float(info.get("pricing_per_1m_input", 0.0)),
+                    float(info.get("pricing_per_1m_output", 0.0)),
+                )
+        except Exception:
+            pass
+        return (0.0, 0.0)
+
     provider_rates = _RATES.get(provider.lower(), {})
     # Exact match first, then prefix match, then wildcard
     if model in provider_rates:

{effgen-0.2.2 → effgen-0.2.3}/effgen/models/anthropic_adapter.py

@@ -33,6 +33,7 @@ from effgen.models.base import (
     ModelType,
     TokenCount,
 )
+from effgen.models.errors import ModelAuthError, ModelNotFoundError
 
 logger = logging.getLogger(__name__)
 
@@ -391,8 +392,15 @@ class AnthropicAdapter(FunctionCallingModel):
                 metadata=metadata,
             )
 
+        except (ModelAuthError, ModelNotFoundError):
+            raise
         except Exception as e:
             logger.error(f"Anthropic API call failed: {e}")
+            msg = str(e)
+            if "401" in msg or "authentication_error" in msg.lower() or "invalid x-api-key" in msg.lower():
+                raise ModelAuthError("anthropic", self.model_name, msg) from e
+            if "404" in msg or "not_found_error" in msg.lower():
+                raise ModelNotFoundError("anthropic", self.model_name, msg) from e
             raise RuntimeError(f"Generation failed: {e}") from e
 
     # ── Generate stream ───────────────────────────────────────────────────
@@ -802,3 +810,23 @@ class AnthropicAdapter(FunctionCallingModel):
         self.total_cost = 0.0
         self.total_tokens = 0
         logger.info("Usage statistics reset")
+
+
+# ---------------------------------------------------------------------------
+# Self-register with the ProviderRegistry on first import (idempotent)
+# ---------------------------------------------------------------------------
+def _register() -> None:
+    try:
+        from effgen.models.anthropic_models import ANTHROPIC_MODELS
+        from effgen.models.registry import ProviderRegistry
+        ProviderRegistry.register(
+            "anthropic",
+            AnthropicAdapter,
+            ANTHROPIC_MODELS,
+            env_keys=["ANTHROPIC_API_KEY"],
+        )
+    except Exception:
+        pass
+
+
+_register()