effgen 0.2.2__tar.gz → 0.2.4__tar.gz

{effgen-0.2.2/effgen.egg-info → effgen-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: effgen
-Version: 0.2.2
+Version: 0.2.4
 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
@@ -127,6 +137,11 @@ Provides-Extra: monitoring
 Requires-Dist: wandb>=0.16.0; extra == "monitoring"
 Requires-Dist: tensorboard>=2.15.0; extra == "monitoring"
 Provides-Extra: all
+Requires-Dist: pytest>=7.4.0; extra == "all"
+Requires-Dist: pytest-asyncio>=0.21.0; extra == "all"
+Requires-Dist: pytest-cov>=4.1.0; extra == "all"
+Requires-Dist: pytest-timeout>=2.2.0; extra == "all"
+Requires-Dist: pytest-forked>=1.6.0; extra == "all"
 Requires-Dist: vllm>=0.2.7; extra == "all"
 Requires-Dist: faiss-cpu>=1.7.4; extra == "all"
 Requires-Dist: chromadb>=0.4.18; extra == "all"
@@ -148,6 +163,11 @@ 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
@@ -194,6 +214,8 @@ Dynamic: requires-python
 
 | | Date | Update |
 |:---:|:---|:---|
+| 🚀 | **14 May 2026** | **v0.2.4 Released**: ModelRouter with CostBased/LatencyBased/FirstAvailable policies, transparent provider failover, cross-process SQLite rate-limit coordination, persistent cost tracker + `effgen cost` dashboard CLI. [See changelog](https://github.com/ctrl-gaurav/effGen/blob/main/CHANGELOG.md#024---2026-05-14) |
+| 🚀 | **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](https://github.com/ctrl-gaurav/effGen/blob/main/CHANGELOG.md#023---2026-05-04) |
 | 🚀 | **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](https://github.com/ctrl-gaurav/effGen/blob/main/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](https://github.com/ctrl-gaurav/effGen/blob/main/CHANGELOG.md#020---2026-04-09) |
 | 🍎 | **8 Apr 2026** | **MLX & Apple Silicon support merged** (PR #4): Native Metal GPU acceleration via MLX & MLX-VLM backends. `pip install effgen[mlx]` |
@@ -392,6 +414,41 @@ Production API<br/>
 
 ---
 
+## 🆕 ModelRouter — Smart Multi-Provider Routing (v0.2.4)
+
+Route requests across 9 cloud providers automatically — pick the cheapest, fastest, or first available:
+
+```python
+from effgen import PolicyBasedRouter, RoutingContext, CostBasedPolicy, LatencyBasedPolicy
+from effgen.models.capabilities import Capability
+
+# Build a router: try fastest first, fall back to cheapest
+router = PolicyBasedRouter(policies=[LatencyBasedPolicy(), CostBasedPolicy()])
+
+ctx = RoutingContext(
+    prompt_tokens_estimate=500,
+    user_budget_usd=0.01,       # stay within $0.01
+    latency_budget_ms=3000,     # need response in under 3s
+    required_capabilities={Capability.chat},
+)
+
+decision = router.route(ctx)
+print(decision.chosen)      # e.g., ProviderModelPair("cerebras", "llama3.1-8b")
+print(decision.eliminated)  # [(pair, reason), ...] — fully explainable
+```
+
+**Transparent failover** — `route_and_execute` retries on rate-limits, 5xx errors, or timeouts and seamlessly moves to the next-best provider.
+
+**Cost dashboard** — track every API call:
+
+```bash
+effgen cost today          # per-provider per-model table
+effgen cost week           # rolling 7-day view
+effgen cost set-budget 1.0 # set $1/day cap
+```
+
+---
+
 ## 🎯 Agent Presets
 
 Get started instantly with ready-to-use agent configurations:

{effgen-0.2.2 → effgen-0.2.4}/README.md

@@ -36,6 +36,8 @@
 
 | | Date | Update |
 |:---:|:---|:---|
+| 🚀 | **14 May 2026** | **v0.2.4 Released**: ModelRouter with CostBased/LatencyBased/FirstAvailable policies, transparent provider failover, cross-process SQLite rate-limit coordination, persistent cost tracker + `effgen cost` dashboard CLI. [See changelog](CHANGELOG.md#024---2026-05-14) |
+| 🚀 | **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 +272,91 @@ Production API<br/>
 
 ---
 
-## 🆕 What's New in v0.2.2
+## 🆕 What's New in v0.2.4
 
 <details open>
-<summary><b>Top 5 features in v0.2.2</b></summary>
+<summary><b>Top 5 features in v0.2.4 — ModelRouter & Cost Optimizer</b></summary>
+
+1. **`PolicyBasedRouter`** — composable routing engine with three built-in policies. Pick the cheapest provider within your budget, the fastest under your SLA, or simply the first available — and combine them freely.
+
+   ```python
+   from effgen import PolicyBasedRouter, RoutingContext, CostBasedPolicy, LatencyBasedPolicy
+   from effgen.models.capabilities import Capability
+
+   router = PolicyBasedRouter(policies=[LatencyBasedPolicy(), CostBasedPolicy()])
+   ctx = RoutingContext(
+       prompt_tokens_estimate=500,
+       user_budget_usd=0.01,
+       latency_budget_ms=3000,
+       required_capabilities={Capability.chat},
+   )
+   decision = router.route(ctx)
+   print(decision.chosen)      # e.g., ProviderModelPair("cerebras", "llama3.1-8b")
+   print(decision.eliminated)  # [(pair, reason), ...] — fully explainable
+   ```
+
+2. **Transparent failover** — `route_and_execute(ctx, fn)` retries on rate-limits / 5xx / timeouts and seamlessly moves to the next-best provider. Each hop fires a `RouterEvent` to registered subscribers.
+
+   ```python
+   from effgen import load_model
+
+   def call_provider(pair):
+       model = load_model(pair.model_id, provider=pair.provider)
+       return model.generate("Hello!").text
+
+   router.subscribe(
+       lambda event: print(
+           f"Failover: {event.from_provider}/{event.from_model} "
+           f"→ {event.to_provider}/{event.to_model}"
+       )
+   )
+   result = router.route_and_execute(ctx, call_provider)
+   ```
+
+3. **Cross-process SQLite rate-limit coordination** — share a single rate-limit budget across multiple workers:
+
+   ```python
+   from effgen import RateLimitCoordinator, SQLiteRateLimitStore
+
+   store = SQLiteRateLimitStore("~/.effgen/rate_limits.sqlite")
+   coordinator = RateLimitCoordinator(storage=store)  # WAL-mode, BEGIN IMMEDIATE
+   ```
+
+4. **Persistent cost tracking + `effgen cost` CLI** — every API call persists to SQLite; query spend instantly:
+
+   ```bash
+   effgen cost today          # per-provider per-model table
+   effgen cost week           # rolling 7-day view
+   effgen cost by-provider    # lifetime totals
+   effgen cost set-budget 1.0 # set $1/day cap (BudgetExceededError at 100%)
+   ```
+
+5. **Fully explainable decisions + budget guard** — `RouterDecision` records every eliminated provider and why (`"rate_limited"`, `"no_key"`, `"cost_exceeds_budget"`, `"latency_exceeds_sla"`). Configure a daily spend cap; the router automatically fails over to a free-tier provider when the budget is hit.
+
+</details>
+
+<details>
+<summary><b>Top 5 features from 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 (and earlier)</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 +669,47 @@ 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-0.2.4}/README_PYPI.md

@@ -37,6 +37,8 @@
 
 | | Date | Update |
 |:---:|:---|:---|
+| 🚀 | **14 May 2026** | **v0.2.4 Released**: ModelRouter with CostBased/LatencyBased/FirstAvailable policies, transparent provider failover, cross-process SQLite rate-limit coordination, persistent cost tracker + `effgen cost` dashboard CLI. [See changelog](https://github.com/ctrl-gaurav/effGen/blob/main/CHANGELOG.md#024---2026-05-14) |
+| 🚀 | **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](https://github.com/ctrl-gaurav/effGen/blob/main/CHANGELOG.md#023---2026-05-04) |
 | 🚀 | **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](https://github.com/ctrl-gaurav/effGen/blob/main/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](https://github.com/ctrl-gaurav/effGen/blob/main/CHANGELOG.md#020---2026-04-09) |
 | 🍎 | **8 Apr 2026** | **MLX & Apple Silicon support merged** (PR #4): Native Metal GPU acceleration via MLX & MLX-VLM backends. `pip install effgen[mlx]` |
@@ -235,6 +237,41 @@ Production API<br/>
 
 ---
 
+## 🆕 ModelRouter — Smart Multi-Provider Routing (v0.2.4)
+
+Route requests across 9 cloud providers automatically — pick the cheapest, fastest, or first available:
+
+```python
+from effgen import PolicyBasedRouter, RoutingContext, CostBasedPolicy, LatencyBasedPolicy
+from effgen.models.capabilities import Capability
+
+# Build a router: try fastest first, fall back to cheapest
+router = PolicyBasedRouter(policies=[LatencyBasedPolicy(), CostBasedPolicy()])
+
+ctx = RoutingContext(
+    prompt_tokens_estimate=500,
+    user_budget_usd=0.01,       # stay within $0.01
+    latency_budget_ms=3000,     # need response in under 3s
+    required_capabilities={Capability.chat},
+)
+
+decision = router.route(ctx)
+print(decision.chosen)      # e.g., ProviderModelPair("cerebras", "llama3.1-8b")
+print(decision.eliminated)  # [(pair, reason), ...] — fully explainable
+```
+
+**Transparent failover** — `route_and_execute` retries on rate-limits, 5xx errors, or timeouts and seamlessly moves to the next-best provider.
+
+**Cost dashboard** — track every API call:
+
+```bash
+effgen cost today          # per-provider per-model table
+effgen cost week           # rolling 7-day view
+effgen cost set-budget 1.0 # set $1/day cap
+```
+
+---
+
 ## 🎯 Agent Presets
 
 Get started instantly with ready-to-use agent configurations:

{effgen-0.2.2 → effgen-0.2.4}/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__ = "0.2.4"
 __author__ = "effGen Team"
 __license__ = "Apache-2.0"
 
@@ -74,30 +74,94 @@ from effgen.models import (
     AnthropicAdapter,
     BaseModel,
     CerebrasAdapter,
+    CostBasedPolicy,
+    CostTracker,
+    FireworksAdapter,
+    FirstAvailablePolicy,
     GeminiAdapter,
     GenerationConfig,
     GenerationResult,
+    GroqAdapter,
+    HFInferenceAdapter,
+    LatencyBasedPolicy,
+    LatencyTracker,
     ModelLoader,
     OpenAIAdapter,
+    PolicyBasedRouter,
+    ProviderModelPair,
+    ReplicateAdapter,
+    RetryPolicy,
+    RouterDecision,
+    RouterEvent,
+    RoutingContext,
+    RoutingPolicy,
+    SQLiteCostStore,
     StreamChunk,
+    TogetherAdapter,
     TransformersEngine,
     VLLMEngine,
     load_model,
 )
 from effgen.models._rate_limit import RateLimitCoordinator, RateLimitExceeded  # noqa: I001
+from effgen.models._rate_limit_store import SQLiteRateLimitStore  # 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 (  # noqa: I001
+    AllCandidatesExhaustedError,
+    AmbiguousModelError,
+    BudgetExceededError,
+    InvalidRequestError,
+    ModelAuthError,
+    ModelNotFoundError,
+    ModelRefusalError,
+    ModelTimeoutError,
+    ModelUnavailableError,
+    NoCandidateWithinBudgetError,
+    ProviderTransientError,
+    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 +268,60 @@ __all__ = [
     "StreamChunk",
     "GeminiAdapter",
     "CerebrasAdapter",
+    "GroqAdapter",
+    "TogetherAdapter",
+    "FireworksAdapter",
+    "ReplicateAdapter",
+    "HFInferenceAdapter",
     "ModelLoader",
     "GenerationConfig",
     "GenerationResult",
+    # Router (v0.2.4+)
+    "PolicyBasedRouter",
+    "RoutingPolicy",
+    "RoutingContext",
+    "RouterDecision",
+    "RouterEvent",
+    "ProviderModelPair",
+    "FirstAvailablePolicy",
+    "CostBasedPolicy",
+    "LatencyBasedPolicy",
+    "RetryPolicy",
+    # Tracking (v0.2.4+)
+    "LatencyTracker",
+    "CostTracker",
+    "SQLiteCostStore",
     "RateLimitCoordinator",
     "RateLimitExceeded",
+    "SQLiteRateLimitStore",
+    # Errors
     "ModelRefusalError",
+    "ModelAuthError",
+    "ModelTimeoutError",
+    "ModelUnavailableError",
+    "ModelNotFoundError",
+    "AmbiguousModelError",
+    "NoCandidateWithinBudgetError",
+    "ToolIncompatibleError",
+    "AllCandidatesExhaustedError",
+    "BudgetExceededError",
+    "ProviderTransientError",
+    "InvalidRequestError",
     "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 +332,37 @@ __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-0.2.4}/effgen/api/__init__.py

@@ -1,6 +1,6 @@
 """effGen API Server v2 — Production Gateway.
 
-Phase 12 modules:
+Modules:
 - openai_compat: OpenAI-compatible /v1/chat/completions and /v1/completions
 - queue: RequestQueue with priority, fair scheduling, backpressure
 - pool: AgentPool with min/max size and auto-scaling

{effgen-0.2.2 → effgen-0.2.4}/effgen/cache/__init__.py

@@ -1,4 +1,4 @@
-"""effGen caching subsystem (Phase 14).
+"""effGen caching subsystem.
 
 Provides prompt-prefix caching and result caching for tools and agents.
 All components are pure-Python and have no required external dependencies.